David speaks with Alex Fefegha, the co-founder and principal director at COMUZI.

This episode was two or three years in the making, and it was definitely worth it.

Alex Fefegha works at the intersection of design, programming, speculative fiction, and art, expressing societal challenges and exposing the technology systems that touch us.

Alongside his work at COMUZI, Lex is an artist currently exploring AI creative's potential via a number of experiments such as the Hip Hop Poetry Bot and Thames Path 2040.

Lex was an associate lecturer at the University of Arts London's creative computing institute, teaching a module on computational futures and artificial intelligence, and holds a Masters degree in Innovation from internationally renowned Art & Design School, Central St Martins.

πŸŽ™ Listen in your favourite podcast player

The Knowledge with David Elikwu - Podcast App Links - Plink
Podcast App smart link to listen, download, and subscribe to The Knowledge with David Elikwu. Click to listen! The Knowledge with David Elikwu by David Elikwu has 29 episodes listed in the Self-Improvement category. Podcast links by Plink.

πŸ“Ή Watch on Youtube

πŸ‘€ Connect with Alex:

Twitter: @lexmakesthings

Website: https://www.comuzi.xyz

πŸ“„ Show notes:

03:20 | Alex’s background

14:47 | Getting into computers through floppy disk culture

16:25 | Getting arrested the night of an exam

17: 32 | Starting performing arts

21: 09 | The flaws of the education system

24:27 | Alex’s first exposure to technology

28:38 | Traveling an unmarked path

31:51 | Creating a record label

36:06 | Community and music

41:58 | Computers are stupid

55:18 | Finishing master's and offered an opportunity to do a PhD

59:00 | Building AI tech

1:00:59 | Breaking the machine

1:07:42 | Getting interested in climate change

1:12:23 | Introducing AI to Hip-hop

1:17:25 | Prompt engineering

1:24:47 | Can AI be truly creative?

1:32:47 | The distinction between human learning and machine learning

1:40:19 | Jobs of the future interface with AI

πŸ—£ Mentioned in the show:

Comuzi | https://www.comuzi.xyz/

Lex GPT | https://lexmakesthings.fun/

Richie Brave | https://twitter.com/RichieBrave

Tooting & Mitcham | https://www.tmunited.org/

Pecan | https://www.pecan.ai/

Audacity | https://www.audacityteam.org/

Bonnet football club | https://www.bonnetbayfc.com/

Wired pr | https://wired-pr.co.uk/

Music X-Ray | https://www.musicxray.com/

Mike McCready | https://twitter.com/mikemccreadypj

A&R's | https://www.careersinmusic.com/what-is-a-r/

SoundCloud | https://soundcloud.com/

Feminist Internet | https://www.feministinternet.com/

Digital One Dead bot | https://www.wionews.com/world/deadbots-the-digital-soul-that-can-speak-for-you-after-your-death-478639

New Inc Museum | https://www.newmuseum.org/

Ross Goodwin | https://rossgoodwin.com/

Es Devlin | https://esdevlin.com/

Rakim the hip hop | https://en.wikipedia.org/wiki/Rakim

The Thames Path 2040 | https://www.newreal.cc/artworks/the-thames-path-2040

MKBHD | https://mkbhd.com/

Chat GPT | https://openai.com/blog/chatgpt/

Full episode transcript below

πŸ‘¨πŸΎβ€πŸ’» About David Elikwu:

David Elikwu FRSA is a serial entrepreneur, strategist, and writer. David is the founder of The Knowledge, a platform helping people think deeper and work smarter.

🐣 Twitter: @Delikwu / @itstheknowledge

🌐 Website: https://www.davidelikwu.com

πŸ“½οΈ Youtube: https://www.youtube.com/davidelikwu

πŸ“Έ Instagram: https://www.instagram.com/delikwu/

πŸ•Ί TikTok: https://www.tiktok.com/@delikwu

πŸŽ™οΈ Podcast: http://plnk.to/theknowledge

πŸ“– EBook: https://delikwu.gumroad.com/l/manual

My Online Course

πŸ–₯️ Career Hyperdrive: https://maven.com/theknowledge/career-hyperdrive

Career Hyperdrive is a live, cohort-based course that helps people find their competitive advantage, gain clarity around their goals and build a future-proof set of mental frameworks so they can live an extraordinary life doing work they love.

The Knowledge

πŸ“© Newsletter: https://theknowledge.io

The Knowledge is a weekly newsletter for people who want to get more out of life. It's full of insights from psychology, philosophy, productivity, and business, all designed to help you think deeper and work smarter.

My Favorite Tools

🎞️ Descript: https://www.descript.com?lmref=alZv3w

πŸ“¨ Convertkit: https://convertkit.com?lmref=ZkJh_w

πŸ”° NordVPN: https://go.nordvpn.net/SH2yr

πŸ’Ή Nutmeg: http://bit.ly/nutmegde

🎧 Audible: https://www.amazon.co.uk/Audible-Free-Trial-Digital-Membership/dp/B00OPA2XFG?tag=davidelikw0ec-21

πŸ“œFull transcript:

Alex Fefegha: [00:00:00] Where when I think of AI, and you know, for me it's a tool that can analyze data and make predictions. It's a good predictor, it's a sophisticated predictor. That's what it does, it makes constantly predictions based on the data you've trained it on, based on the task that you've set it, it has no consciousness of what the heck is doing. It doesn't understand the impact to what it's doing. It just generates something, generates the result.

And so for me, I think as I've worked with technology over the years and predominantly focused on AI, like I've always seen that computers are stupid.

David Elikwu: Hey, I'm David Elikwu. And this is The Knowledge. A podcast for anyone looking to think deeper and work smarter. In every episode I speak with makers, thinkers, and innovators to help you get more out of life.

This This week I'm speaking with Alex Fefegha, the co-founder and [00:01:00] principal director at COMUZI. This episode was two or three years in the making and it was definitely worth it.

Alex and I talked about his incredible journey from being a young budding footballer to becoming a IT technician, to eventually being able to work with huge AI projects with companies like Google. And so we dug into, you know, this idea of being a creative technologist. What does that mean? We talked about AI, how AI tools have been used and could be used in the future.

And we talked about the incredible story that I actually didn't know having known Alex for years, but I didn't know the full story behind COMUZI, which is the creative studio that he has built with his co-founders. So it's a really incredible story, and I know you're gonna love this episode.

You can get the full show notes, transcript and read my newsletter @theknowledge.io.

Every week, I share some of the best tools, ideas, and frameworks that I come across from business psychology, philosophy and productivity. So if you want the best that I have to share, you can get that in the newsletter at [00:02:00] theknowledge.io.

You can find Alex online on Twitter @lexmakesthings. And if you love this episode, please do share it with a friend. And don't forget to leave a review wherever you listen to podcasts because it helps us tremendously to reach other people just like you.

So I would love to get your thoughts on a lot of this GPT-3 stuff, everyone's talking about AI right now. But you were like my original AI Jedi because, funny enough, so there's this thing that happens on Twitter where, every time a new topic becomes popular, everyone becomes an expert on it. So suddenly everyone is coming out with all their threads, all their think pieces, all their blog posts, and they're presenting themselves as an expert on whether it's AI, whether it was Ukraine, whether whatever it is. Everyone's an expert, but you from even before the pandemic from years ago, were always the person that I would come to when I had questions about AI. Cuz you have actually been working on AI for longer than pretty much anyone that I know. [00:03:00] You've done projects with Google, you've done projects with other companies as well. So I would love to, I mean maybe, maybe we can start with your background. I think you call yourself a creative technologist and you've been in technology for a while, but was that something you grew up with? Did you always have this, passion for technology or was that something that you had to develop kind of as you grew older?

Alex Fefegha: Yeah, it's something that I actually mentioned on my website. So I have a website called, it is entitled lexmakesthings.fun and it's called Lex GPT, where you can chat with me. I originally made it in 2019 and then I thought maybe chatbots are boring and then I saw in recent hype that chatbots are the in thing again. So I brought my chatbot back. But in that story I actually talk about how like I was not that done who was focused on tech and stuff, like my Headspace was all about sports, like I wanted to play soccer or football. Now I feel like I've been [00:04:00] speaking to too much Americans about football too much, I just said soccer. Oh my God. Yeah, I wanted to play football. I wanted to play Centerback, you know, I was big, tall, fast, had all the attributes to be a really good center half. And my focus was literally can I make it in football? If I can't, can I make it in athletics? Cause I was fast. If I can't, can I go be like a sports psychologist or a strength and conditioning coach?

One of those two. I was really interested in performance and what helped someone perform better. Like, you know, and that, and, and that element of helping someone perform better kind of brings me eventually as we talk about the AI conversation. Cause that same thing really was something I really focused on. I was really interested in performance. And that inspired me from like when I was like 15 because it's, you from South London, you know the economy at home is tight. So are you getting the nutritions you need? Are you getting the rest you need? Are you getting the care you need? How [00:05:00] does that accommodate your body to be able to perform at a level, especially if you are a young athlete trying to get to the highest of your sport? What are the environment you need in order to be performing very well, especially at the time that you know, you might talk so much about the physical attributes of an individual, but the mental attributes of an individual was something that wasn't spoken about. And so I was very interested in sports psychology in terms of its impact with golf, with people who played golf.

And for me, who knew nothing about golf, I was like, what do you mean you need sports psychologists for this, da da da da da. But now as I'm older and I watch sports more and myself playing sports, when I understood about, I would perform in environments well where I felt confident, but in environments where I didn't feel confident, I'd second guess myself. Things wouldn't feel natural, I would all of these particular things. So my background very much was sports. Like I studied a Sports Science, [00:06:00] BTEC, and then added in, like I got like personal trainer qualifications, sports coaching qualifications. I was actually planning to go to university to study sports science as a BA and then somehow get a scholarship to go to the states to study some sort of masters in like sports psychology or something like that. Or try to navigate the American education system where you have to study first and then you get drafted into some sort of major league or something. Like I proper, that was my whole headspace. The only thing that actually stopped me going back was actually like not having money. Because you're here, I didn't really understand like the division one, division two, division three education system that the states has, you know, every young athlete in the states want to go to division one university. That's where the full scholarships are gonna be, that's where trying to get into the NFL, trying to get into the MLS, trying to get into the NBA, trying to get into NHL and the other sort of sports [00:07:00] association leagues or you know, individual sports division one had the facilities and stuff. I didn't know anything. I just wanted to be there. But there was a lack of money because I wasn't getting four scholarships. It was like, you are basically, the plan was you play, like, you have like, these companies where they tell you to play exhibition matches. So you might be one player, you're playing us a bunch of players, and then you have like people who watch you and they scout you and then they help you go to the states, but you need money for a certain process. Then you need a university to watch you. Then you need a university to keep a scholarship, it ain't gonna be a full scholarship. And all of those stuff, you know, and, and, and now you can kind of see it in like 2023. There's a bit more networks.

Like, you know, for example, I have a interest in American football. It was something that people were trying to get me to play for years and, you know, couple years ago I took it up as a sport after. I was in [00:08:00] America in the gym and someone saying to me like, yo, you know, what's your 40 yard dash? And I'm like, I don't know. And he's like, bro, you can go to the NFL and make money, da da da da da. And then I happen to be in the gym in the UK. Someone comes up to me again and goes, what sports do you play? And I'm like, Gym. And he's like, nah, I think you should play American football, da da da da. So I took it up and I took an interest and I began to realize, okay, there's like an NFL academy, which now actually helps young players from the UK be able to have the opportunity to train, live the American football lifestyle, play American football matches, but also have the exposure to those division one universities and other universities which can help move their destiny in. And that's a whole completely different game and so, I guess at the time when I turned 19, I was very much like asking myself like, I haven't got a professional club contract. I can have a play semi-pro football [00:09:00] and really try to do my best to like perform and pray that I can get an opportunity. And, and, and you know, from playing professionally, I didn't mind what level I played. I just wanted to play football professionally. I really loved it. I felt like I was good at it. But there's many other people who are just as good at me or even better than me trying to pursue the sport. And you know, and it's such a hard sport. There are people I played with that I thought they were unbelievable and they ain't pursuing ball. They're not playing it right now. And so it's just the hard thing.

So, you know, one of my closest friends, Rich, was studying more media technology at Brunel and it's actually happens to be called Digital Design now at Brunel and he was learning how to make websites. He wasn't the best with it as I banter him, but you know, I began to like be interested in the tech world, mostly around the culture of tech startups. Like, can you [00:10:00] create your own startup? Can you, you know, you're hearing stuff about like, Larry Ellison the founder of Oracle buying a beach, and you are watching this on E-News. There used to be three of us, me, Richard, and Ernest. And we'd be in Ernest's house watching these videos and stuff and we're like, Yo, I love to be a tech entrepreneur, da da da da. And then it was like, how do we enter it? How do we dabble in it? So at the same time, just that nature of who we are, we were very entrepreneurial. Richard was always the person that was like an eBay seller. Like, he would sell his phone on eBay. I remember in College, he would sell his phone, stack the money, save, buy more stuff, and then buy a phone back after, that was rich. That was Rich's personality. Where the good thing I had is I knew a lot of people, so it was like, how do we combine that mindset and that skillset to basically try to make legitimate [00:11:00] ways of making money?

And I grew up in Peckham, South London, in the 'Nam in the estate. And in the Nis, as we call it, you know, so my headspace, my environment was, if you need to make bread before the ages of 18, bread equals money for those who heard me speak, you know, you do what you need to do, the creative entrepreneurship, you know, and that type of stuff and you did that. That was, that was the model. I already knew that was the model, like pre 18, like, you know, and like I said, the economy was tight. You wanted new clothes, you wanted new trainers, you wanted to eat, you wanted to experience stuff. And those who had to do shiny stuff were the creative entrepreneurs, I can't say that I was tempted to dabble in it. I guess I just had that fear of like getting caught and having a whole different path. So for us it was like very much in our headspace was like, how can we like, try have the same energy [00:12:00] that these men have on a more legitimate scale? That was literally the whole mindset like, I had no grand plan whatsoever. I was just broke and just wanted to make bread. Like, I had no, oh yeah, there's a grand master plan and I knew about technology. I knew that AI was coming, I had no clue.

David Elikwu: Yeah. So I was just gonna ask, did you have much access to computers and stuff at the time? Because I remember, so I was talking to Richie Brave about this, if you know him. But what I find funny is that I remember there was a, there's a distinct period of my life, probably even when I was quite young, like 14, 15 or so, where I had access to computers, but only through the school library and I lived across the road from a library. So I saw books about coding and so I, there's some stuff that I learned, but I learned it for the purpose of designing websites. But beyond that, I had no idea that there was an actual industry of legitimate tech companies and I think I just didn't even see that at all. All I saw was, okay, [00:13:00] here's ways you can make money today, and that's about it. And so when it wasn't useful for that, then I didn't really take it any further beyond that.

Alex Fefegha: Now, I think it is a very true point. You mentioned the library, like the first my grandad, you know, who was a University lecturer. He predominantly lectured in Nigeria, but I think he decided to move to England back to like, to wanna study. I mean, not study here, to lecture here. And he had a PC that he bought, set up, but he didn't know to use it. So he was going to like, there was a charity, there was stall a charity called Pecan, which was all about like, I think it's an employment support charity, but at the time in 2003 in that they had like learned how to use computers. So when my Grandad predominantly come from world of typewriters and, you know, writing, you know, pen and paper and then move into the computer, it was, he had that first exposure to learn in that way.

I think I, when I was much more younger, I think I did have a PC for myself. But then I went, [00:14:00] my grandparents wanted me to study in Nigeria. So I studied in Nigeria for like two years or something when I was young. And I remember the IT lessons, you have one PC, they bring one computer to the class between 50 people then fam.

You watch the teacher teach how to use the computer fam. One, if the student's lucky that day, it's your lucky day to type on the computer. So I remember that and then when I came back, had the PC at home, but no internet, so I'm just playing games on there. But I used to go to the library and when I would go to the library, I would like get into the floppy disk culture. So I would meet people who had games on floppy disks. And then what I would do would like go to these websites, get my floppy disc, download like Mario, Prince of Persia, Bad Games, and put it on a floppy disc, and then go home and put it in. Cause I didn't have, there was no internet fam, so you need to get them games on floppy [00:15:00] disc. I remember playing mini clip, mini clip games, like, ah, that was my whole library experience was gaming. I literally just gamed in the library for a long period of time and then in secondary school we used to have ICT classes and ICT classes, you know, the teachers did their best, you know, I think ICT classes were classes where most of us did not really pay attention. It was kind of a free for all lesson to basically just like play games. I think I played games throughout every ICT lesson and watch football compilations on YouTube at the time of like, different players, man. Now then, I didn't even know 480p or whatever quality that it would probably be. But back then, all I'm doing is watching YouTube compilations of footballers, watching Namar, Santos and Gantzel and all of them, stuff like that.

Like that was my life. Like I did not, you know, and then what happened was [00:16:00] actually, there was a moment in like 2011 where I actually got kicked out of college for, unfortunate story. you know, I got arrested the night of an exam. So why I studied at BTEC for Science, I did an A levels in performing arts at the same time. I went to a college in Sunbury called St. Pauls, and I was actually playing for a team. Who was I playing for? Tooting & Mitcham, like, they had like they're semi-pro club, but they had like this academy program. You know, the whole thing when you are in that ages, you would have football clubs over professional, mostly semi-professional football clubs or like the lower league professional clubs who would have these academy systems set up around football. So you would have kids who would come and study, I'm a BTEC and then you would like get the football teams kit, you would train for the team and you would play on Wednesdays in like the youth leagues and stuff like that. So I went [00:17:00] there and I got arrested and then they were like, yeah, you gonna have to repeat college again. And I was like, damn.

One day, the person I was seeing at the time in 2011, wanted to go to, I was looking at college viewing days. Cause they were at one college and they wanted to change the college. So I went to view the college and then I was, started performing arts as A level. And that is hard cuz they're like two different worlds of sports science and performing arts. But because I did a Drama GCSE, it was like, I have an opportunity to do an A level and when you are trying to pursue the sports sort of thing, it's very rigid in terms of like, if you're somebody who may be, has been raised in a very academic way. Like my granddad was very big on academics for me. So you're not really having a lot of leeway in terms of like, if you wanted to study A levels, it's gonna be a madness. But if you studied a BTEC in sports science, it was, the course was structured around your football so [00:18:00] you can train and do a lot of things around that. So I was like, okay, you know, I'm a personal multiple interest. That's always been me. I want, I'm gonna keep this A level in performing arts. And so there was performance, you're doing a performance at this period of time, there was a lunchtime performance and it was an evening performance. I'm trying to get back.

So I happened to go through Kings Cross, she then mentions that she doesn't want to come anymore. So I go, okay, cool. I need to cut. So I'm going to start leaving by my once, a police officer goes like, can we, you know, stop and search you? And I go, No, because like I need to get to college. Like, I can't, I'm running late now. I need to speed. And growing up in ends, I know how to navigate the stop and searches, you know, where some of my friends would give the police, you know, a hard time and that type stuff. My thing I realized over time was if I just spoke to them in a nice way and said, [00:19:00] yeah, you are right. Duh, duh had a bit of banter. The searches is less vigorous, it's more blessed and everybody's your friend and they actually leave you alone. Like, that's something I learned was it's like, it's interesting. It's like they wanted you to provoke them and I learned that the tactic is not to provoke in any way, shape, or form. And so like, I said to police officer, look, I'm rushing. I need to get to like college, like can you just allow me to go? The police officer goes, you are smelling of weed. Dude, at that time I had never smoked weed in my life. So I was like, ah, please kind I go? He's like, okay, I'm gonna call back up. Call back up. I thought it's just gonna be a thing of, look, I'm bigger and taller than everybody, so I acknowledge and understand that, you know, that I am physically intimidating. So,[00:20:00] I thought call him back up, so you can do your search and you basically just let me go that basically, you know. Alright, thank you mate. Off you go. No, I actually got arrested. I got arrested, I got cautioned for the servant. What's it like disturbing the public peace? Something like that, like, you know, I got arrested in Kings Cross station and handcuffs and all that stuff. I think, you know, I remember the solicitor or that I had at the time basically was calling out the police officer for abusing their power and yeah, that messed up cuz I meant to be at performance and you know, it was a play and I had been a very key part to that play.

And yeah, so, but the fortunate thing about that, which leads to this part is I, you know, obviously growing up in secondary school, your whole mindset was the way how things was, and this is maybe the flaws of education system is, you know, it's a game, right? Get you know, GCSEs [00:21:00] don't get three years in college, try to do college in two, finish University in three. Like everything's timed. And I think at the time when they told me that I'm gonna have to do first year again, I was so angry and I was like, what do you mean duh dare so, I need to find something that worked out for me. And I began to look at apprenticeships cuz I had this window for a period of time, of how do I study and make money. And I think my ego was crushed that I would have to study again. So I looked for apprenticeships and at the time, this is one year after the Tories have taken power and I know the Tories' whole agenda was to try to improve vocational qualifications and, apprenticeships and stuff at the time. So the apprenticeship program I think, had always been there. I'm not sure, we had launched in a new way. So this is early stage apprenticeship program and this is also a recovering economic market as well after the global crisis. So I'm applying for apprenticeships. I [00:22:00] think I applied for like an Office Admin 1, I applied for an Accounting 1 and I applied for an IT 1. The Accounting 1 was sweet cause it was paying like four bills a week, 400 pounds a week for those who are not familiar with that, 400 pounds a week in 2011 was, I don't even know what I'd done with that money if I got it, you know, but it was funny that one, because you had people who had graduated from university who were actually applying for the same apprenticeship program as you.

So that's why I mentioned the economic crisis because you had folk who you would think would be in a graduate scheme, would be doing that were actually applying for the same apprenticeships. They were willing to go to college to study at the same time you've just graduated, you got a whole degree and your hair competing with me for this opportunity and stuff. But it was the IT apprenticeship that accepted me in a primary school in South London in like Clapham Junction somewhere. They weren't the ones who hired me. It was a company who hired me and then [00:23:00] I got put in the, in the primary school as like a, it's like a staff member, basically as a contractor. But the company who hired me was dodgy. I didn't get paid for three months,

David Elikwu: Wow. Really?

Alex Fefegha: Yeah, and then one time when they tried to pay me, I think HSSBC thought it was fraud, so they closed my HSSBC account. And then I got given, at first they gave me a check to set a check balance is fraud, duh duh. Then I got money in the award of cash after Three months, which was really nice. I blew the cash, I don't know what I did with the cash, but I did what anyone who would do in 2011. How designed 2011? I was 17. I did what any 17 year old would do with money in 2011. Maybe, actually, no, maybe some 17 year olds might be different, but you know. Yeah, I had 1,333 pounds. I was out there, man. I was up.

David Elikwu: It goes a long way in ends.

Alex Fefegha: Yeah, it goes a great way. But the apprenticeship was my first exposure to technology [00:24:00] because I spent a lot of time fixing and building stuff. I had to fix the computers. A lot of the times, the broken keyboards, the all of these things, putting 'em together, putting a lot of parts together, taking up computers apart, trying to fix stuff. I spent a lot of time on YouTube, watching YouTube videos to learn how to do stuff, how to learn, how to use different software. This is a world when the cloud did not exist, well it probably, the cloud did exist, but this is a world where people went migrating and stuff to the cloud. So the server was still in the office right next to you in a big black heavy box. This is the server. You gotta make sure the server and it's a backup server. You gonna make the sure that stuff has backup all the time. And that stuff crashes, it's game over for you. That was my world, that was like the IT apprentice. And then there was a lady in the primary school who was like the IT manager and she played like other roles, but she was like, I was like her protege and then that's when I learned how to like start being comfortable with the world of like code and stuff [00:25:00] was because I had waffled and said I can make websites when I couldn't. So I had to learn very much on the job. I had to learn how to come up with new ways to like hack my things together. And that's where the hack energy kind of came from because one of the first things I ever did was, there was audiobooks that the school wanted to basically, I mean audio cassettes that they had and they wanted to basically upload them to the internet, or upload them to like this shared folder so that kids could listen to them and study without you having these like cassette players. And so that was the first challenge I ever had. And I had to almost like hack my way to do so, and that was like me. It was a very simple hack, I literally just played, I think I learned how to play the cassette, get the cassette player, connect it to a computer, record it in Audacity. [00:26:00] Audacity, for those who don't know, was a recording software. If you ever were spiting and making music back in the days, you might put a beat, you put Audacity on there and you record. Like, I was a music man, manmade music. Like, and that's the next step of the journey that comes into it. And that was literally like my first time of like being told to come up with something and hacking your way together with technology. I didn't code anything, but it was the ability to have a solution at the period of time, understand what technology or what softwares, what things particularly exist, and almost like come up with a solution. And that was a, you know, a very early stage example of no code movement in a way and that type of stuff. Then it intrigued my interest a lot further to want to explore it. However, what happened as a football team came back and said to me, hey hey Alex, do you want to come join us? I think at the time I'd even tried for Bonnet football club who were in League two at the [00:27:00] time. But I was so outshake and everything. After working an office job for the first time of my life and drinking coffee and eating cakes that I performed terribly, so I had to get back in shape, in time,

David Elikwu: Just a quick question.

Alex Fefegha: Yeah, go for it.

David Elikwu: I'm interested to know, cuz you have now started work early and you are kind of now on a different path from a lot of your contemporaries. A lot of the people that you grew up with, you are working people are still at school at this point in time. Like people are still going to college, people are still going to school. People haven't even gone to university yet. You are already working, drinking coffee, going to an office every day. How did that feel and how did you adapt to that? Almost a completely different life. Because also, I think you, you mentioned specifically that this was like a very recent initiative, I remember this as well, like they'd only just started bringing back apprenticeships. And so before that point, those opportunities didn't even exist. So now you are on, you're kind of traveling an unmarked path. I'm [00:28:00] sure even if you looked a year or two above you, I don't even think anyone before you would've had that same opportunity to do that in the first place.

Alex Fefegha: I think it holds you to my personality. I think I was one of those people who was very much, I liked it. I think that's why I did it. I didn't have to be trapped in the forms of education. Like, you know, I think it was if it wasn't because I was trying to pursue, pursue trying to be an, an athlete and having a career around sports, I didn't really mind the aspects of like working, cuz then I was like, okay cool. I was probably trying to build this whole career plan. How do I work in IT? How do I get my qualifications? Like you need all this CompTIA A+, CompTIA ++, CompTIA. I remember some guy kept talking about CompTIA, CompTIA that's he kept saying CompTIA. I just kept hearing so much for him like, and I was thinking, alright cool, how'd I get my CompTIA this, CompTIA that. I remember at the time I was doing so well in the school that the school were [00:29:00] proper thinking about how can I be a, like a proper employee there and things like that. Like I was proper, I enjoyed my time. The only issue is I wasn't getting paid on time and it wasn't a school's fault cuz they weren't the people who employed me, but I was enjoying my time. Like, I was like IT person, but then between before nine o'clock, I'm the receptionist man, you know, I gotta pick up the calls, I gotta hand in, you know, I gotta help, you know, parents, I gotta do this, I gotta do that. At 12 or 12:30, when the time lunchtime is, I'm a lunchtime helper. Like that was it, you know? And the dinner ladies, they looked after me. Well, you know, they fed me cause I was broke. So I was looked after quite well and supported really well. I made good friends with some of the teachers, like I would play Sunday League football with one of the teachers at the time and played for the Sunday League team.

Like, so things for me was like a really, it was actually a really cool wooden experience. I didn't really feel like, ah, [00:30:00] I'm missing like out on friends and things like that. I was like, I see my friends after, but I'm like, I'm getting paid and I'm working. And that was like, that was cool for me. But I obviously knew that I still wanted to pursue sports, so I kind of quit the apprenticeship to go play for sports. Cause my football team was like, yeah, we want you to come play. So I went to pursue that. And then kind of saw after a while that, okay, cool, I don't see myself getting a professional contract. I need to basically start thinking about other stuff. And so Richard was studying at Brunel at the time, and this goes back to music. And one of the things we had, there was three of us Madjeck, Madjeck was somebody I went to secondary school with and Madjeck was making music in Polish. So he was rapping in Polish and things. And we would like collaborate and it's like a song with, myself, Madjeck and actually Flohio. Flohio, she's doing bits right now. She's a rapper and she's bloody amazing, like I [00:31:00] love Flohio so much. and the name of the track was actually called Pure School's Records. I'm still trying to find that track. I might have to message Madjeck and ask him, yo, where that track at?

David Elikwu: Was it in Polish?

Alex Fefegha: I don't know if he Rap in Polish or English, but it was like that type of,

David Elikwu: Yeah.

Alex Fefegha: Yeah, those eras was cool man. That was like some full creative energy stuff and I had a couple friends who rap and things. We decided to create a record label called Pure School's Records and a recording studio. And Pure School records was, I had a bunch of friends who made music and they were really cool and we thought, could we sign them? And like, could we help them, help their careers grow? And it was like, how do we make money? And there was two things, make websites and that's what we did. Cause Richard was trying to make a website. He didn't really know to make it. Our friend Ernest had a friend called Gabs who was studying computer science in Luton, Gabs then put us on how to make websites. And from there I just ran with it. Like it was like, cause at the time I had really kind of gone out court. I'm not gonna pursue sports like [00:32:00] that from a professional capacity cuz I'm not sure if I'm gonna really be able to have the opportunity. I'm just, whatever I put myself into, I'm just really gonna, that same athlete mentality. You wake up, you grind, you focus, du,du,du,du. It's the same thing I'm going to do. And so, you know, getting exposed to making sites, I was like, boom, I'm gonna make sites. I'm gonna learn how to make this stuff. I'm gonna learn how to make this thing even more creative than the other person. Da da da da da. And that became my obsession, like for a period of years.

And then we tried to, then we realized, I think we got like a startup loan. And then money from like making sites. And then we tried to fund an artist from Manchester who actually had been getting like a lot of listeners and plays on radio. And he had a really interesting single, and the single sounded so good. It got like, record of the week and like Charlie Sloth's BB1 Extra. And we paid, I think we only sold 27 singles. [00:33:00] Something crazy, it was something nuts, something nuts. Like we put 6,000 pounds into a whole, we had like radio plugin, we had stylists, we had a video shoot where you got the stylist, you get the Davinci's, you get this, you have the model, you have the studio, you have the RED camera. It was shot on the RED camera, bro. Like at the time that thing was not cheap, fam, you get a RED camera to shoot.

David Elikwu: Even now it's expensive.

Alex Fefegha: This is 2013, bro. 2013. I am 19, bro. Like, and this is what we were trying to do, radio pluggin. Had a radio plug, I think she's really doing one. I think it's Wired pr. I might be so wrong, but her name was called Rachel. She was amazing. No, she was doing PR. We had a radio plug and she was doing PR like press PR and stuff. Sold 27 singles and then we were like, nah, they saying it, man. But at the same time I was very much inspired by Troy Carter [00:34:00] and Scooter Braun. Troy Carter was the manager of Lady Gaga and Scooter Braun is the manager of Justin Bieber. And at the time they were very much talking about, like, I think there was a thing at the time when celebrities began to invest a lot in social media startups and other startups. And they were very much talking about how technology was gonna transform the music industry. And I began to grow a lot more closely in it. I began to start buying into what they were saying, and then I was like, I wanted to be involved. Like I wanna, I want to do that stuff. I want to, like get technology and music industry together. Because at the time, this is different now where Spotify streams, YouTube streams, all of that stuff contributes to somebody's charting position. At the time, the record labels hated Tech York destroying our business or eating ourselves, and so I became an advocate for that. And what actually happened was the first week of university where I meant to go to Uni, I ended up in [00:35:00] Barcelona to go to the musical tech conference 2013. There was a company called Music X-Ray, which is founded by a guy called Mike McCready. He had message me in LinkedIn saying, I like your thoughts and perceptions. Cause what I was doing at that time was I was spamming everybody's LinkedIn to the point that LinkedIn blocked me from adding people because people are like, who the heck are you? They don't know you.

David Elikwu: What were you sending them?

Alex Fefegha: Like, Hey, I'm Alex Fefegha da duh. Like, work with us and we can help your music artists growing in their like outreach with their fans.

David Elikwu: Okay. So were you branding this as like a agency, like a full agency already.

Alex Fefegha: Yeah, that's where community came from. And I'm gonna explain where community, the business community comes from, it comes from the term is Community and Music. And it was trying to how do we basically develop, how do you develop communities around fans? It was playing around with the fan club concept. How do you develop like stronger connection between a music artist and their fans? You know, many musical [00:36:00] artists for every Drake who can stream millions, make millions. There's artists who struggled to make money and it was like if you had a artist and you had a hundred very close friends and you were able to maybe, for example, generate a thousand pounds of value from each fan, in some way. That could be a mixture concerts, merchandising, you know, connection, building, whatever you could do that's a hundred K. And how much music artists can say, they make a hundred K from their art. And so we had that sort of passion, like could we somehow generate door relationships and stuff? And this is kinda like a P Patreon aspect cuz Patreon was launched to sort of help creators and music artists by paying monthly to support them in some way. So that was our, our whole headspace. So at least she went round to every record label, every music agency at [00:37:00] first, I went walking around with like, with like paper and like throwing in and being like, Hey, can you give this in like a proper, I didn't know the game, I just went everywhere, bro. My whole strategy for years has been, if you don't know me, the whole world's gonna know me. And that was my energy. It's been my energy throughout the whole journey, even of like community where it is now comes from those same things. And like I said, I stopped playing ball. So whatever I was gonna do now I need to make this thing work. That was my mentality, that was my head space. I had dropped out of Uni, so I was, I was hungry fam in these streets.

So I like literally was messaging people on LinkedIn, you know, trying to get them to understand that I think I was going to like music meetups, and basically saying the Big free isn't Sony, Universal and Emi, it's like the big free is Spotify, Instagram, da da da da da. Like that was [00:38:00] heavily my focus. And then went to the Future music conference, which Mike McCreedy was a founder, a music x-ray was sponsoring and it's like I was planning to, plan Freshers Week. I just got my first student finance payment. So I changed my mind, I decided to go to University of West London cause I had a London College of Music and I was actually gonna study music management. Like, I pivoted. I made a full life pivot in like one month prior to all of these things. And then you know, fortunately the University of West London had a scholarship program and I was grateful cause I did really well in college. So that with the UCAS points things, so I got like a, I actually got the first year of Uni funded, through the scholarship that they had a chancellor's scholarship. So I was good, I was like, fam, you know, I didn't get the full student finance pay, like, you know, you get like a loan and you get like a maintenance loan. Just due to certain circumstances, I only had the one level. I was like, ah, cool. I'm calm. So the pees have dropped. I'm thinking, yeah, I'm gonna stack this pees boy.[00:39:00]

Then I get an email saying, Hey Alex, da da da da. I met Mike McCreedy before, a couple months before. And you know, we talked about music technology. So Music X-Ray, that's the first time I heard of AI. Music x-Ray is basically, I dunno if he still does this now, but it was a platform where music artists put, or modern music artists could send their songs to this platform and pay, but the difference compared to this platform compared to others was music x-Ray had built a lot of relationships with A & R's that they would get guaranteed listeners. Cause I think that's the thing that every musical artist, when they, record labels used to have like these SoundClouds and all these places where it used to be like, Hey, send your music for your SoundCloud. Send your music to this email. Or some record labels have no contact details whatsoever. Go to a music Mac night. And then the A&R's had power but A&R's had this great gravity of power around music. And so Music X-Ray's [00:40:00] purpose was like, you, the budding artist, you send your song to platform because Music X-Ray has this relationship to A&R's you get a guaranteed listen.

But what Music X-Ray was doing, which was really cool, and this is why I first began to understand AI, was they were basically trying to find out what made it hit single, what makes a hit song. And so the data set they were creating was a, A&R's opinions on like looking at a song and being able to go, this is a hit song, this is not a hit song. These are the reasons why it's not a hit song. Being able to have that database of songs and then somehow trying to create this algorithm, this sort of artificially intelligent, you know, I hate the word artificial intelligence to be fair, but a machine learning algorithm, which I should say that somehow could have the ability to predict what a hit song sounds like, or what it feels like or what it needs to come together.

So it was essentially trying [00:41:00] to build like this really smart tool that could help, like record labels know how to piece a hit song together by using some sort of AI. I use the term machine learning because I don't think computers are intelligent. I think they're very stupid. But I think artificial intelligence is the word that folks use and it's good for marketable reasons, but I like to use the word machine.

David Elikwu: Wait, do you mind expanding on that. Why do you think computers are stupid? Just quickly before you go back to the story.

Alex Fefegha: Cause computers are not that great, like, I think they're great at being able to analyze data and stuff that is the programmed artificial intelligence. It can produce and it can analyze, it can crunch, it can absorb mass amount of data at a very fast scale. Like very fast pace, very at scale, you know. But when you working with this tech for so long, like it's not that great. You know, [00:42:00] it's ofcourse something intelligent, right? It's an assumption of this autonomous being, this is a being who's autonomous, has agency who has a consciousness of what it does and how it uses intelligence to influence the scenario. That is what intelligence looks like, if you were to ask someone what is intelligent or not, it's a problem solving aspect. But from this autonomous, you can put, you know, where when I think of AI, and you know, for me it's a tool that can analyze data and make predictions. It's a good predictor, it's a sophisticated predictor. That's what it does, it makes constantly predictions based on the data you've trained it on, based on the task that you've set it, it has no consciousness of what the heck is doing. It doesn't understand the impact to what it's doing. It just generates something, generates the result.

And so for me, I think, you know, as I've worked with technology over the years and predominantly focused on AI, [00:43:00] like I've always seen that computers are stupid. And I really for a long time have been wanting to do a lot of work around, like computers are stupid. Like at pre covid, one of the things I'm trying to do with a bunch of University of Arts London students who studied advertising was to create an exhibition at London College of Communication, which is part of the University of Arts London, a exhibition on artificial stupidity. And what we were going to show was examples of how computers more stupid and like, but it was to create this humorous, dark humor aspect of it. Because for me, I think, throughout my whole journey when I first got immersed into the AI space and I, I, you know, I would just wrap up the COMUZI story.

You know, I run a company called Comuzi with two of my closest friends are Kon, Richard. Comuzi started from trying to bring music artists and fans together thru technology. We decided to create our own App called the Comuzi App and we wanted to basically [00:44:00] reward communities that were built around particular music artists. It was inspired by Lady Gaga's Little Monster. Lady Gaga used to have a social network called Little Monster. I didn't really know how much of the users on there were actually real, but it was basically Lady Gaga universe in a way. And we were very inspired about how we can capture that and bring that around other artists.

One of the issues at the time was we were good at, we've always been good at building stuff, you know, because we were hackers, we were hacked something together. We'll make something together. We read all the books on building an MVP and all of these things. But the issue was you would go to investors and investors would say, how do you bring a return on investment in three years? And in our heads we're like, you know, I think for us, when we've been so inspired about trying to build things that nobody else was doing, and looking maybe to America, as our source of some inspiration, we knew that some of the things we were trying to build were gonna take a number of years to have like, success and penetration. Cause you're asking to change [00:45:00] people's behaviors. And if you're asking for us to bring a return in free ads, I'm not sure we can do that. You know, unless you scale really fast. And we literally, we did not have the language to understand what the heck they were saying. So what happened very early, we tried to explore different ideas. We tried to build music tools, we tried to build small business tools, we tried to build healthcare tools. And the same issue happened throughout all the time. So we decided to say, you know what, cool for a couple of years, let's not build our own ideas. Let's build ideas for other people. And then that became a thing for like, from like 2015 all the way to now has kind of been building stuff for other people to be fair. AI comes to me around 2017 when I began to focus on it a lot more, and that was originally about, I actually don't have a bachelor's degree, but I managed to get into a master's to study in innovation at University of Arts London. I go into that master's partially because I had a portfolio of work and I had my mind was set on free [00:46:00] Universities, Brunel, RCA, and Central St. Martins, which is a college as part of University of London. And I heard of it cause Kanye West wanted to go there and he couldn't get in. And so my whole plan for university was literally, I don't have a bachelor's degree, but I have a ton of work. And I have references that can back me up. I've got my Headspace. It was a very arrogant approach to doing it and yeah, I got into a master's and then in that master's I began to be interested in artificial intelligence. And what I was interested originally was about, I was exploring the ethical implications of, you know, AI and originally about the internet things. Then I got bored cause the internet things I thought was boring and there's been a lot of conversation about having smart objects and smart homes and devices in your house and I thought, okay, let's just explore artificial intelligence. And so the focus there began to grow on, you know, AI's grown society and I began to look at it from the aspect of race [00:47:00] and gender. So I looked at AI and policing, in the legal system in America where, you know, there was an AI tool that was being used to assess if folks were most likely to re-offend or not. And a bunch of investigative journalists for the publication ProPublica was able to sort of do this investigation where they saw that this AI tool must, was basically saying a bunch of black folk were more likely to re-offend than those of other races who their criminal records were actually a bit more seasoned. There was seasoned criminals versus first time offenders, and the first time offenders are being considered to most likely to be able to re-offend. So I was very interested in exploring that because as I was looking at from a technology perspective, I think there was a lot of conversation about ethics and AI, but they're mostly done by academics. And I've mentally always struggled with fury. You know, I'm a very practical person. I'm very much like a maker, [00:48:00] and I was interested in like all this ethical conversation that's going on. All these theoretical aspects that's happening on this high level hair. How do you take these furies and turn them into like implementable stuff that a designer or a developer who work with these technologies can actually embody into the design and the development of their software or these tools that are powered by AI.

And so that was my interest and so I began to look at this particular re-offending tool. I was trying to understand what was the issue, the issue was the data set. You know, the data set was fed on, you know, let's make up it, the criminal justice system isn't like the most, you know, loveliest system dedicated to black folk in the world, right? So you, you have that issue. The second thing is they had a questionnaire which basically asked folk, like, when you get arrested questions like,[00:49:00] how many people in your family's been arrested? You know, how many of your friends have gone to prison? Like it tries to build this profile on you without maybe looking at going sensitive, yeah, Are you black? Are you white? Are you, you know, from a Hispanic background? Are you from this, from that? No, it asks you particular questions, which can create this demographic information and try to produce that paints a particular picture. But just what I say I call it, it's not intelligent, right? If it's intelligent, you would know. AI should be able to tell, okay, cool. Yes, this person might come from, like, if I use me as an example, which I use many times in the past, if you had to ask me some of these questions there, I could answer those questions and say, yeah, you know, there are friends who have been to prison. They're this, they're that, this is my environment, these are my background but does that make me most likely to re-offend? You know? And the intelligent aspect comes from being able to have that humanness, right. [00:50:00] Where you got this tool and it's able to be able to go, whoa, okay, this person seems like they have, you know, they've refunded, there's some context maybe why they've had this thing. Somebody stolen a bike. There's probably a context behind that, you know if I say this person is most likely to reoffend, am I putting them through a system of I'm not really, in a prison doesn't really re habitate folks or rehelp them, get back into society really well, it doesn't, you know, am I now gonna put them through this cycle of in and out, in and out of jail? Or do I go first, I'm offending, how do we implement systems or frameworks in the real world that will help this person be able to live their life and continue to be, you know, grow to be amazing and whatever they want to be in that particular way, rather than putting them through this sort of [00:51:00] chaos and, you know, system and stuff like that. And obviously that's a song goes, isn't at the road of the human, but then I'm like, okay, this is why if humans can't even answer that question well, what makes you think that AI is gonna be able to answer that properly? And so I used to have the saying that, AI just makes that bad human decisions faster.

That is the aspect and the difference. And so I began to focus a lot there, and then I actually stopped. And the reason why I stopped was black guy talking about racial bias, talking about gender bias in AI systems. What folks were doing was ignoring everything I was talking about, you know. I've got blogs on the internet, right? Writing about design implementations, how companies can take design frameworks or approaches, how companies can care about these things. Like I'm a designer, I'm a maker, I make stuff, I don't know anything about like, diversity and inclusion and you know, disregard to those who do, like, they're people who study and they know all their stuff. Like, they can break this down, they [00:52:00] can speak it. They know the furies, the frameworks, the tools, the everything. I don't know these. I don't know all that stuff, I'm trying to make sure I don't swear, but I don't know that stuff, you know, and thought was happening is. I'm on panels, yeah. And I'm talking about these things. This one's coming to ask me questions about like diversity and inclusion. So how do we implement diversity and inclusion to that? Or like, I remember once having a company asked me to do diversity and inclusion workshops and I was like rahh. And I didn't, and this is what I'm saying I have, when I described this, I'm not trying to devalue the importance of those diversity and inclusion and, you know, the efforts. There are many people I know who work in that space who are amazing and they actually do really great relatable work. But I'm not that person. That ain't my world. That's not my expertise. Go to the experts and I would have to find people who are experts and be like, this is the person, work with them. That ain't my expertise. Like, I'm trying to be seen as a technologist. I'm trying to be seen as a designer. I'm trying to be seen as a [00:53:00] researcher. I'm trying to be seen as those things, as somebody who has like the ability to work with this AI models, create my own, create my own data sets, produce these things. And so I had to move away. Like I stopped talking about it, to the point where now there are people that still ask me, like, museums and all these places, Hey, let's come do this documentary, come do this thing. Can you speak about this? And I go, no, I refuse it. Because like, I felt like I was being devalued for what I cared about. So I decided to switch, and that brought me to the world of AI and creativity. And that's how I got into the generative AI space, which is now what is called, I don't know what it was called then. I used called creative AI space at the time, that's called generative AI. Moon was called generative AI before, and I never knew.

I decided to get into the space of looking at how, and this goes back to the early vision of me about looking at performance. And I was interested in like, how can we use AI to help folks be more creative? And how do you explore that? [00:54:00] What does creativity even mean anyway? You know, from an academic standpoint, there's so much different definitions of creativity. When does creativity actually happen? Is that in here? Is that out here? When does it actually happen? Especially we don't know so much of the human brain. We still don't know where creativity comes from. We don't even know how the brain functions in that way. Is it from here? Is it from here? Where does it happen? We don't know. You know? So you're trying to get a computer to replicate human creativity or do it better. That is the whole purpose of artificial intelligence. Get it, do the same thing as humans or possibly better. Like, how's it gonna do that when we don't even know where creativity happens? And I saw that as like a very interesting pursuit of exploration and opportunity.

When I finished my masters, I had been offered an opportunity to do a PhD. I applied for a PhD, I didn't get it. But I was still very much inspired to explore artificial intelligence [00:55:00] and creativity. And so for a period of time I had to, similar thing. I had to build my reputation, get the respect of folks in order to do stuff. So I did a lot of work around AI. Like one of the first projects I ever did was around like, how do you create a feminist Alexa? And that was in response to like, at the time you had this voice assistance where if you spoke to them in a sexual manner or in a, you know, very rude, crude manner, they weren't designed with any prompts to tell this user, Hey, don't basically speak to me in this way, shape, or form. And so the first, you know, there was an organization called Feminist Internet, which spun out University of West London. And they're very much inspired. And they were like, we wanna somehow build these technical prototypes around like that shows how we build better voice assistance. And so that was the first third project I had done around AI. In the context of like that particular [00:56:00] side and as a creative technologist, that title simply comes as a so title saw in America. It's like, you are not a software engineer, you are not a out and out designer. You're not out and out of any of these things. You are all those things. You are journalists. You might have a strength in certain area, but you are kind of somebody who, you know, brings those two worlds together, of that creativity and that technology. And I always see the role as like creative explorations of technology. And that was this particular project, you know, I taught, you know, about 20, 20 students, 40 students taught about like 40 students, how to make their own voice assistants. And then I prototyped, I think it was like eight voice assistants that could, they 40 divided by eight is five. So they had like five groups who had late. Like five. Yeah, my math is all over the place right now. But yeah, there was like eight voice assistants that I had to basically make and prototype and that project went, went really very well [00:57:00] actually, to be fair. And the folks began to like who's this kid? Who's this guy? Duh, duh. Worked on a project that was talking about how do you make a explainable algorithms, so how can algorithms basically explain to you it's thought process and stuff like that in terms of, like why it's made a particular decision. Like how do you design that so that somebody else can understand that and looked at that from healthcare mental health perspective. They've built a lot of like different chatbots and things. Looked at how we could use chatbots in like employment opportunities and jobs and stuff of like local government. Then I got an email basically saying like, come to the U.S come and talk about AI and art. And I think at the time I was like, why am I being spoken about an AI and Art? I think I had done a project with an artist group where it was called the Digital and Dead Bot, it's a Comuzi site and the digital and dead bot basically has a conversation me about death. [00:58:00] And it captures your stories about death and the whole purpose is, it captures your stories to tell another person that story, to capture their stories about, and having this relationship about death in a way. So a lot of the sort of chatbots I was making was more like this explorational aspect of dust and technology for the sake of, for me, it was cool, it was fun. I had wanted to do an AI and art residency and I didn't get it and I got asked to mentor the artists on the residency. So I was like, oh, okay, this a catch 22.

And then, you know, I ended up building the tech as well. So it was a really fun, cool, experience. And then I went to, America, New York for the first time, April, 2019. Yeah, April, 2019. And I spoke at the New Inc Museum, about AI art and authorship, and I went there to speak about like the role of AI and where does AI support in the creative process. What happened in the audience is there was, you know, from Google AI, there's a team called Google [00:59:00] Artists and Machine Intelligence where they had worked with artists in the past to do a lot of work on artificial intelligence and looking at how, you know, artists are wonderful interrogators. They question the technology, they do stuff that show the technology in a different way, especially when you take away the commercial side away from it. And I spent a lot of time really living in the commercial. And I really wanted to live more on the artistic side, and so I knew they were gonna be there. So I knew I had to shine and I tried to navigate the space really well, and I met a bunch of folks. I met Ross Goodwin. Ross Goodwin is a technologist, a creative technologist who had in a past at created algorithms around script writing. So he wrote a film with another gentleman called Sunspring, where they basically made a film where the script is written by an AI. He had done a lot of work with Es Devlin, the stage designer, set designer as well and she's amazing. Sat one of my creative heroes and he had [01:00:00] created a lot of algorithms to some of her work with AI and expressions of AI and creativity around like poem generation and stuff like that. In a, I'm bringing the audience to take a part in generating the poem. And then I met him in New York and I spoke to him about my idea of like, I know that these AI, the databases for these sort of poetic generation tools are 19th century poetry. But I'm interested in breaking the machine. And this goes back to my desire of showing how stupid the computers are.

I'm interested in breaking the machine. I want to train it on rap. Rap isn't, you know, structured English as you know, predominantly a lot of these databases, data sets are trained on. I wanna break the machine to see what we can create with it. And for me that was, I always struggled to describe the project, which grew into what is called a hiphop poetry button now. And my interest was just, I just wanted to break the machine and see what happens if you train the AI on rap. Does it stop it from working [01:01:00] or does it create a whole new chaotic madness in a way, but also is why I wanted to study. I went and rap to be seen with the same poetic appreciation compared to like maybes, like other genres and things like that. So the whole idea at first was to create a rap club where you train, you create this algorithm which generates, lyrics in a teleprompter, and then you have an artist performed with these lyrics in real time. And it was just an exploration, it was just meant to be fun, play around, see what we can do. And then Google supported it and then the project began to evolve into verse. Could we train an AI model to rhyme? And not just during rhyming, not the last word, but having particular rhyming schemes like how Rakim the hip hop, you know, the rapper would have and all these things. What could we do to create a level of sophistication in a way, not just on the last [01:02:00] line? Could it have the context? Could it, you know, because when rhyming for me, when I say intelligence rhyme, to be a good rhymer, if you're a rhyming in a rap battle in an environment, that is context, that is wittiness, that is being able to understand how to feel the audience reaction, imagine, think five steps ahead, what words you're gonna say you know, to, I don't know, violate your opponent, get the audience reaction. And that's a very, you gotta think about that at a very fast time.

A machine learning model doesn't really have any knowledge of that. It doesn't know how to do that because it ain't intelligent, it ain't got context.

David Elikwu: Hmm, Yeah.

Alex Fefegha: You know, it's good at analyzing data in the case of this particular project, I had created a data set from scratch, where I literally had to think in my head, who are all the different music artists I can think of who are good at [01:03:00] telling stories with rap, like they are able to describe a scenario, an event, a moment, descriptively, you know, because when I downloaded data sets of the internet and fed it to a machine learning model and also machine learning model to generate text, and literally she just went, N, N, N, N, N word. N word. N word, B,B,B,B,B,B,B.

David Elikwu: I think I remember this, I remember towards the start when you were trying, it was with an existing data set and it was coming out with like racist stuff, sexist stuff. I guess it's because it's drawing on the context of the time at which the original text is coming from.

Alex Fefegha: Yeah, and I always say The Internet's Repository of evil, that so. See it like this, you got this machine learning models that are trained on the internet. You ask it to generate texts, you feel it with a data set that's already inflammatory in a way. What's it gonna do? Free for all inflammatory, inflammatory, inflammatory. So we were [01:04:00] like, and it was like, okay, cool. I'm gonna create a data stuff from scratch where I'm really going to look at every artist lyrics and I'm gonna only put the songs in the data set that the artist has very reduced swear word as much as possible, but they're just hella descriptive. Like, you know, they describe the podcast and, I don't know, two brothers speaking on a mic, living in this life, trying to get time. I don't know. Something, something but it was literally like you describe a podcast as like some next level intellectual conversation of the mind, merging of minds, like describing it in a breaking down way. Because I needed the ML model when it was gonna be given a prompt. I needed it to generate lyrical content that was very descriptive and had a storytelling aspect to it. Because over time we were interested in playing around with this model or [01:05:00] this concept of an oracle and it was something that I flirted around with and I didn't really, I think in the end, I tried to stay away from it, but at first one of the ideas was this aspects of an oracle where it examines the path of predictive future. And so when Covid hit and it was like, okay, maybe the rap glove idea don't really work cause we're not allowed out. But maybe if we created this, like, you know, the world is a bit sad right now and people are down. So we created this bot that you can ask it a question every day based on, you know, different categories and it would give you some sort of affirmation every day in the form of rap. Because, you know, for me, rap is an art form that has empowered me. It's a, rap form that's made me reflect on life. It's an art form that I've utilized to just say, look, I just wanna have a good time. I don't wanna think about the world right now and that type of stuff. And how could we capture that and use this [01:06:00] AI tool? For me, it was an experiment. I was just interested in experimenting, I had no grand picture for anything, it was just necessarily an experiment and that became the hiphop poetry bot. And this is pre, what the hype you have now?

GPT-3 did not exist, no chat G P T existed. This is chat. This is GPT-2, using GPT-2 at this point, to basically try to make this work and things like that. And, you know, now fast forward, you've got a range of different stuff in its space. You know, I recently was working with the University of Edinburgh and Edinburgh Future Institute to create a project, which is still ongoing. And I'm trying to think about what's the best way to present this to the world. It's called The Thames Path 2040. And it's inspired by the neural I can never pronounce this word, observatory.

David Elikwu: Oh, observatory.

Alex Fefegha: If you got a better grasp in, yes. Hey, my English [01:07:00] is some of words are too bad fam. It's the game. You know what I'm saying? but yeah, but with the new realm, they had commissioned a bunch of artists, including myself, um, to basically, look, you know, we are interested in climate change, climate data. We also wanna make AI more legible to an audience. But whole idea was like, could you take climate change data? Could you look at a context in the future? And based on, you know, if someone was to click on a particular date in the future at a particular location and give them a scenario, what does the world look like at that period of time? So, you know, majority of, a large percentage of London is built on the flood plain, and that's closed back all the way from the Roman times and then how London grew as a city along the river and those particular things. And it's built on the flood plain. And we saw a lot in 2021 where places like Stratford [01:08:00] had a lot of flesh flooding and those particular things. There's also the story of the great North Sea flood where it actually affected parts of London, Amsterdam, Essex, it did a lot of damage. And that what led to the Thames barrier being created, where I live right now in Woolwich I live right next to the Thames barrier. I see it every day from my house.

And so I was interested in, you know, 2040 NASA sort of data predicts that London would be on the flooding in 2040, 2040 is not a long time. It's 17 years away. My math is correct. And so I wanted to somehow create this experience with a Gant model. And Gant models are what you see right now, currently powering things like mid journey, stable diffusion, Dao E and I can't remember if there's any others. And so the whole idea was that somebody [01:09:00] could come pick a date, pick a scenario, and then this AI model generates some sort of visual representation of London in 2040. And what I decided to do was focus a lot on like council states and, because I grew up in a council estate and showing how like particular communities would be affected by, climate change in London. And ideally you wanna do it across the world and stuff, but you know, for the sake of this project, it was built around London. The whole purpose of the project was to show this to policymakers rather than, an audience of Londoners, cuz I think one of the challenging things about talking about the future, it's not scamming people. I don't know if folks need to be scammed on with a blunt or burden misinformation. I think policymakers were the ones who needed to think about this. We ain't talking about this, we ain't talking about London and what's London's provision for London right now? Right now it's still very much [01:10:00] dependent on the Thames barrier. The Thames barrier is meant to be out of commission like in a couple years time and now they're saying no, we can extend it for a couple more years, it'll be fine. But we haven't really spoken about like, flooding in London and things like that. It's interesting and the reason why I wanna bring this up and, and I guess I'll let you speak and you can ask me some questions cause I realize I've just rambled and spoken for time, but I hope this has been cool. When I started on the project, Dall-E, Mid Journey, Stable Diffusion, any other sort of like generative AI tools that we now have in the industry, what did not exist. So what happened was these tools come out at the same time we're working on this project and I guess what I'd done in my, you know, I recently gave a talk in Edinburgh, I think it was in November, I gave this talk in Edinburgh actually. And when I gave this talk, I basically talked about the project and talked about what happened when all these tools got released into the market. At the same time, [01:11:00] I decided to use these tools to see how each sort of Dao E, Mid Journey, Stable Diffusion also the platform we were given, which is the neural platform by the University of Edinburgh Features Institute, and some of my own sort of AI model work, you know, doing the same things but creating my own and trying to see how each model generated the image. Cuz they all generated the image quite differently, you know, and the prompts I'd put in was like, you know, flooding and Peckham, South London and da da da da. And what would happen is where my GANT models purpose would, were very artistic.

I don't know if I share my screen on here. Is it worth sharing it or is it stuff you can put pictures in after? Cause I think it would be good. I think even hiphop poetry bot adding those images in would be good.

But what you could see was, for me, [01:12:00] I was really trying to make these pictures for like artwork. Like you can print them off and you could put them on a canvas piece and it can actually exist in your home. Like I was very much interested in the texture and the layers and the look and the feel of it all. Cause a lot of my images that I had first created myself were just embeddedness of like water, but doesn't look like water, council estates and mixtured in London buses mixtured in this mirage of water. Like, it actually feels like I've actually water painted these things. Where with Dall E, Mid Journey, Stable Diffusion and stuff, my challenge on my issue was, you know, it's, I come from a school of design fiction, speculative fiction, where, you know, speculative fiction and speculative design is about developing tools that are not made for commercial purposes. Their purpose a lot of the times is to engage interrogation of technology in the world [01:13:00] around us and so you create software, you create technologies that they don't really have any commercial value, but they allow us to basically, critically, analyse and critique the world around us and things like that. And one of the things I tried to do was when you speculate about the future, was not trying to make it too real. I wanted it to be ambiguous and very vague so that people don't have to like, go, okay, I need to make this too realistic. But they can just sit in the artistic nature of the particular project with Stable Diffusion, Dao E, Mid Journey and these others, their purpose, obviously as tools is to generate real representations of this thing. So when I'm typing in these prompts, it's giving me, flood, it's giving me water and building high flooding, that type of stuff. And then I'm like, Hey, I'm like, this don't look like Peckham. What the heck? Where did he get these images from? Where did he get this data from? Where did he even like? And that's when I [01:14:00] began to go for me as a creator, as an artist, why I think those tools are cool, why I maybe not sold in these tools for me as a person is I like to have a lot more control of the data set.

Because if I have more control of the data set, I'm able to tell you your story that's very unique to this artwork that I'm creating, like I captured pictures of Peckham, I captured pictures of flooding in London in order to generate, to create this artwork. You know, so I know Peckham, I captured it. I don't know if these other tools in their data sets, there's a Peckham, because Peckham, when you type in Peckham from an image perspective on Google, still has a lot of his heritage from a Peckham where the media represented it predominantly through crime. Not through, you know, the shifts of the, you know, Peckham and all of that particular stuff. So what images have you got on Peckham, right? Is it ones with predominantly [01:15:00] like the, I don't know, police, blue tape and stuff like that. What are those particular stuff? And I was very much interested in like if I can capture images of Peckham, like my council estate, my area, my this, my that, all of these different moments, and I have a large data set of that and then maybe where I might struggle with showing flooding and councilors state or like, you know, most of the pictures or images of flooding are mostly in rural towns. And you don't probably predominantly see them in cities like London. You might see them in cities, in other countries across the world. And how do you bring that together so you can create, so you can feed that term ML model whose purpose is to kind of take this data set and create this immersion of the two. The other platforms do it, but they don't do it in a way, I wanted to do it. And that means I might have to be a good prompt writer to make it work, but I just want to be able to control the data set. And I think those tools have values. It's kind of similar to [01:16:00] like, if you're creating a website builder and with a website builder, you have, you know, Wix, Squarespace and the others, and you can make a website, you can, what you see is what you get, edit, add, you can, you know, put it together and the folks who want to use code and they want to be able to produce their own, maybe control a lot more of the elements and those things. And I've always been on this side of how I've control their elements and stuff in space. So for me, a lot right now is like, even with chat GPT, I think it's cool. I'm very interested in being able to have access to train it on my own data sets to see what I can do with it and that type of stuff.

David Elikwu: I was gonna ask, what do you think of this idea of prompts. Just from what you were saying that I was quite interested in is a lot of people are talking about this idea of prompt engineering. At least that's the phrase people are using. I wonder how you feel about that phrase, but just this idea that, you know, being able to write the correct prompt to get the [01:17:00] answers that you are looking for is becoming almost an art in itself. And if you write bad prompts, you get bad results and so it's actually becoming, I think people are even selling prompts. People are selling the ability to create a complex prompt that gives you the correct answer. And I wonder how, as AI advances, so we're now at a point, and maybe you can fill in the gap of the difference between going from GPT-2, which is what some of your earlier models were based on to GPT-3, GPT-3.5, which is kind of what we're talking about now and actually GPT-4, which might be coming out soon. I think we've seen such a massive shift in the capability where now, you know, I haven't tried it, I haven't tried the chat GPT with poems or anything, but, you know, maybe it is able to write text, it's able to write blog articles.

And so what's happening is people are learning to write incredibly complex prompts so that you can get some of these really cool [01:18:00] end results. I think there was quite recently someone that was on the cover of, I wanna say it was a magazine like Vogue. It might not be Vogue specifically, but on the cover of a magazine. And they designed this magazine cover entirely using AI. I think it was like female astronauts or something like that. But I am really interested in, people have always talked about this idea of, oh, robots are gonna replace people's jobs. But I think it's only now we're finally getting to a point where robots are or AI is not just duplicating text or regurgitating text, but it's also able to replicate art to an extent also. And so there's this interesting paradigm where this magazine came out and people were complaining because, oh, you could have got a real photographer or real artist to design this cover. Instead you've just got a prompt engineer to tap out some prompts, and suddenly you've got an AI generated cover. And I wonder where that leads, where people increasingly maybe start [01:19:00] writing aided by AI or doing art aided by AI. And then there becomes a distinction in what does the creative aspect mean? Does that still count as creativity if you are just being creative in the prompt that you are using, because the amount of maybe talent that it takes to then now be able to create something that looks like a work by Leonardo da Vinci is very different to the level of talent that Leonardo Da Vinci actually had. And so I wonder what you think that means for, I guess the future of creativity and the future of how we might be able to use a lot of these AI tools in our work.

Alex Fefegha: I mean, the first thing, prompt engineering is something that I've just recently started paying more attention to, only because I'm trying to play around with GPT-3 in terms of trying to understand it for writing, just trying to play around with it to see it for us. How good is it in helping people write, [01:20:00] you know, coherent scripts like for film and stuff? Cuz I have a friend who's a script writer, he's very much interested in exploring this area a bit more.

And so I'm trying to play around a bit and you know, been playing around a prompt edit engineering. Cuz I'm trying to understand like, for example, how much of a prompt you need to give to the model. This is what I'm saying, right? The AI is not, it's not intelligent. If you have to do all these sophisticated aspects to get a very good reply, then it's not really intelligent. Right? Or someone could argue is right and prompts the best interface to ask the AI to do a task for us. Is there something we're missing? Is there, should we have a bit more question? Should it be a question based thing? Should we remove away the prompt. I think it's something from a UX perspective and an experience perspective that one has to question. Because if we are expecting everybody to interact with AI, it's [01:21:00] gonna be sophisticated or having a grasp with the English language, then that's gonna be a very big problem.

So ain't that already causes the barrier to entry for accessibility. Cause I think a prompt is a really good, interesting opportunity. It's a good starting point where somebody can get something to be generated and you don't have to go through the process of what I've done in a, which was go through this whole model training aspect and understand Python and all of these particular things there. And with a prompt you can generate something. So I think there's an opportunity there. I think when you talk about the creative process, it is a very interesting debate, it is a very interesting thing.

When I first began to look at AI and creativity, and I was beginning to explore it because I taught a module on computational features in AI, for a couple years at University of Arts, London's Creative Computing Institute. And one of the things I was always trying to do was always have these [01:22:00] questions and these debates with my students at the time around what role should AI play in the creative process? I think for me it is an interesting one because there's two sides of the argument always, and I think we have to have both sides. There is this desire and technology to constantly always be innovating, to constantly, always be producing something new. It's the obsession, I think maybe is it because technology is the closest thing we have to magic in the western secular world that we have this obsession of always the new, you know, I've built my career with existing and always the new. So it's an interesting, interesting place. I think for me, my interest in the space was always about building tools to help creators be more creative. Are we currently doing that at this period of time? Possibly. You know, but I think it goes back to that thing of like, what is creativity? What is a, if you David yourself was to look at someone and say, this person's more creative than another person. How do you judge that? If everybody's creating the [01:23:00] same artwork as Da Vinci, do you judge that person that's super creative or do you say that person's copying da Vinci? Or is it when somebody goes far and beyond Da Vinci that someone goes, oh my God, this person is more creative than Da Vinci. I don't know. I'm just trying to, the way how we, we use, we judge or we analyze creativity in general when we describe people. I think that's why I start off is like, how does someone describe someone to be more creative than another person? If you saw a musical artist and you said, this music artist is more creative, Kendrick Lamar's more creative than Drake, what do you judge that by? Is it the way how his beats sounds? Is it the way how Kendrick make music videos are? Is it the way how da da da da? What is creativity about going against the norm? What is, like all of those particular things where one can truly produce true creativity or you know, and it even goes back where people say, no creativity is original. Nothing's original, it's just remix and replay from others.

So [01:24:00] right now it's a new space. It's like when photography was introduced into the world, there were people who hated the camera. There were fine artists, traditional artists who made artworks and they painted the world. And now somebody produced a camera where you take a picture that captures the field, that used the paint and that caused upheaval, that caused frustration, that caused annoyance. But it also democratized access for those as well, because traditionally painting as an industry was one that was only given to those who either had access to wealth, to have the materials to paint, the materials to paint at the time are very expensive. Now you have a camera and all you have to do is buy film and capture pictures. And then that happened and that created a shift and we adapted to that and now we've obviously had other stuff over the years that's coming to the world of art, you know, for me, I always treated AI the same way as I treat Photoshop [01:25:00] as a space for me to create this design or to create this, to edit this picture. But you know, to create this design, to create this tool, to create this thing, it's AI trying to now become the next Figma or the next something. Is it about embedding AI to help us be more smarter? I think it's trying to understand, like, I think for me, I'm always like asking myself what is AI's role in society? And I think there's a conversation about what is AI's role in the creative process, cuz and I think that's the questions. I think, you know, there are, debates where people go chat GPT should not exist. There's a lot of like different opinions that folks have on this space.

David Elikwu: But I was gonna say, I really love the point that you made, which seemed really poignant to me is that analogy of it to photography, which I think I hadn't thought of it in that particular way. And it makes me wonder if the definition of [01:26:00] what creativity is, is something akin to, you know, or what defines creativity is being able to see the room for creative expression in a tool. Because just like you said, I think actually, and I think it applies equally to painting, I'm sure when people first started using ink dyes or paints, the traditional applications to which those were used are different from what painting then became. It was the people that, you know, applied creativity to those tools that suddenly you're seeing, wow, I didn't even know it was possible to paint in this way. And so we pushed the limits of even how we could draw human faces and human expressions. We pushed the boundaries of that really, really far. And then photography, I think the same way, if you look at some of the very first pictures, a lot of them look kind of similar. People are just, because the camera's just on a big tripod. You hit the shutter button, it's under this hood with the flash and you know, they're all quite similar. And then you have photographers that really push the boundaries [01:27:00] of what we think about when we say photography and what we expect and we keep pushing and we keep pushing those boundaries.

And so I wonder I think because we're so early, and maybe a lot of people have not yet conceptualized how far we could push creativity with AI to know like where the fields that creators will play in. Because right now it's like just on the edge of what we normally do. And so it doesn't seem that creative because all we have is the raw tool which is added on top of, you know, the standard stuff that people do now, the traditional paradigm. And so I think it might be the case that once we can sufficiently develop the current AI tools that we have, you know, creatives will show us the way in a sense, and we will actually then be able to see how creative it is possible to be with the tool like this. And it might be case that in fact, all the things that we're seeing now that seem complex and that seem cool are actually really basic. And it [01:28:00] will be the equivalent of, even if you look at Canva, I think someone made this point not too long ago, which I thought was great, which is that Canva democratize access to Photoshop essentially to, image manipulation and to being able to put together cool graphics. But there's so many graphics that are still rubbish, right? Because being able to access Canva doesn't mean that you have design skills, and it doesn't mean that you actually have a good sense of design. It means that it's a lot easier to get a very basic version of something. I think that's the stage we're at now. I don't know how far we can push it, but it seems like it's a similar paradigm in that with the AI tools, you can get a very basic version with very basic skills. And I wonder if you have a lot more advanced skills, or you apply creativity to that skill set, then maybe there is, we can't even yet imagine how far you could take that.

Alex Fefegha: Yeah. But I think one of the challenging things as well though, cuz when we talked about this analogy and then I heard this conversation, I've been sitting [01:29:00] there thinking I have to give a but and where's my but gonna come from? And then it reminded me that I guess with photography, with Canva and all these places, there's still that human being who is basically the one who is going to this tool and having to navigate the tool to produce this output with AI, it is being trained on data sets of previous created aspects of work, and now being function in its way. You have companies who are selling commercial AI tools, training on the work of artists who are not being compensated for their work, right.

So it's like why I do see the photography analogy and I see it in that way. I have to also look at the side of the artist who, their work are being trained on these AI tools. They're not given the opportunity to be credited. They're not being compensated. And then you [01:30:00] have companies sending these tools and charging you for API access and all of these things. And has to be a way of like, you know, how this is done in an ethical manner. I think you know, democratizing access as an opportunity. But what happens, obviously in our tech driven world, right, is we very much, you know, in a capitalist society, which is not fair, we will push out the small fry the small fry in this case are our artists whose work are not being credited. You know, they're struggling to keep their lights on and stuff and then a tech company comes to say, Hey, come use our AI tool to generate artwork and da da da, da. And so I think,

David Elikwu: Yeah, oh a quick question on that.

Alex Fefegha: Go for it.

David Elikwu: So one thing I wanted to get your thoughts on is the distinction between, for example, I might consider myself a photographer. I am a photographer. I've had photos that I've taken that have been on the Calvin Klein website, I've done fashion weeks, I've done loads of stuff, travel photography, [01:31:00] lots of different contexts. But that is based on a data set that I have built in my mind of all the photos that I've ever seen. And I'm not necessarily giving anyone credit, but I'm learning and I'm building on the data set, anytime I see someone great that is taking pictures, and I look at how did they get this angle, how did they get the lighting, how did they shoot this shot? And I'm kind of reconstructing all of that in my mind and subconsciously, every time I then go and take photos in the future, I'm thinking, okay, what examples have I seen of how people constructed the lighting in a scene? So when I wanna shoot my scene, I am drawing on this data set that I've developed of, okay, I don't remember who did what and how, which particular example I'm copying. I just know that I've seen something like this before and I can do something in such and such a way. And so my photography develops and it grows and it gets better. And so the reason I'm at the stage I am at now is because of all these people that I've learned from in the [01:32:00] past.

And so I'm interested to get your thoughts on what is the meaningful distinction there between, I guess like human learning and machine learning in the sense that human artists have learned from a bunch of human artists in the past and don't necessarily credit them unless they are intentionally trying to copy their work. But if they have just happened to learn from them in aid of creating something that they think is new, it's not the case that they have to credit or they have to pay all the people that they learn from in their journey.

Alex Fefegha: You pose a very good question. I think the issue is you pose a very wonderful question. I mean, the challenge is, is because A. We have to look at where you are looking at a dataset, these AI models machine learning, you know, you are sourcing these dataset on the internet. Predominantly where most of these models, I think there's Common Crawl which is one of the popular data sets that folks use. And then there's another one I remember, isn't Marques Brown Lee?

David Elikwu: MKBHD.

Alex Fefegha: I might get it wrong. Oh my God, I [01:33:00] might get that wrong, you know, YouTuber where he had a recent YouTube episode on creative AI and he showed an example about, there are some examples of AI artworks where you can actually see the artist's signature scribbled in the bottom left. So this is an artist that the AI model has been trained on and it's generating an image. And you can see the scribble of the artist's name on the right bottom corner. So it creates this weird, space. I think the challenging thing why we, it is a different conversation is you've got these tools, they're being trained on the internet, then they're being fed into this model. And then this sort of creativity is being packaged into some sort of tool that's being sold to folks. And I think it's about how we approach copyright law.

That's the first bit, like in terms of you are downloading these things without the consent, and it's like, okay, [01:34:00] if artists consented for their work to be used for the purpose of training, that's quite different. Cause even with the hiphop poetry bot, like the reason why it was reduced in a limited way than it being reduced in this way, that's like I wanted it been an internet system forever, was because of the data set aspect. You know, originally I trained this on data sets of artists on record labels. And you know, because I was working with Google, there was, how you, it's different if I was an independent artist and I'm not doing it for commercial purposes than cool, but you know, because you're working with Google, someone can say that's a commercial reason and they could try to pursue legal action in that particular way. So what I did eventually in the end was actually asked people to send me their data sets and I crowdsource data set. So people, there was an email, there's an email on the hiphop poetry book website where people could send their emails and you can use that to feed the model. And that became this crowdsource data set that I produced in the end for the particular [01:35:00] projects. And even now, if I wanted to go back and work in it, I would have ask a record label. Hey, gimme what your artists, or gimme these artists, please gimme consent. There is not commercial purpose or I'll use the crowdsource data set. But there's a consent aspect in that aspect, I think with a lot of use of these artworks is you are not, there's a difference of looking at a piece of art and going, oh yeah, there's this artist work I looked at and I really liked it and inspired my headspace and I'm gonna take a picture.

But most of the time these AI tools are just replicating the same artwork as another artwork that an artist has done before. We haven't really, I'm not really seeing a distinguishing of the difference. And that's why I talk about the controlling of the data set. For sound like myself, I can control the data set, I can make myself very distinctive maybe. But I think it's that thing of that aspect of where is the data set coming from? It's coming from the internet, it's being taken off particular websites where some of these artists probably do have copyright restrictions and then it's being trained. So you have to address that part as much. I love the question you [01:36:00] asked and I love the question you posed cause I think it's wonderful and what you described, but I think because of that, where is the data set being sourced from? If you had consent from all these artists, then it's fine. But if there's no consent and it's being used, it almost feels like a deceptive way. I think that might be worth even having, it's like if you're interested in exploring this topic more, even having an artist on board who's maybe is work or, you know, something that I haven't maybe paid a lot of attention to be fair myself, like in delving into like the artists who feel like their artwork is being trained and being used in these models. And I know there are now some AI platforms that I think are arts and artists, if they wanna opt out, they can opt out their artwork for being used to train. But I think it's that part, I think it is that part. I think if you've downloaded your art, if you downloaded your data set from a common license, Creative Commons, sort of data's com, I don't know, artwork license, then I think [01:37:00] then you're good. I think you should be fine. It is such a contentious space and I don't have the answers and I personally don't want to have the answers just yet because I think it's a space where needs a lot of discussion and understanding and I think for me, I'm trying to also understand like the space, I'm trying to understand like how did my own efforts contribute to maybe a tech solutionism aspect of the space. So, you know, I'm trying to navigate the space and I'm trying to figure it out for myself, and I'm trying to do this in a ethical way. Cause I know, I know a lot about the ethical communications of this space, and so if I try to avoid it, then it's not nice is it, my brain won't allow me to avoid it. So I'm in that space where I'm literally just trying to like, read all the arguments, read all the perspectives. I definitely do find it cringey when I see like, random people on Twitter go, Chat GPT's gonna change everything [01:38:00] and da da duh. That's a bit where I'm like, God, that's so cringy, man.

But in terms of like everything else, it's an interesting space. I'm very interested in generative AI in terms of healthcare though, that's like a big thing for me. How can it help, like medical professionals and things like that. You know, prompt engineering, could prompt engineering, if you thought medical professionals had to prompt engineer and gave it Chat GPT on all the other tools, what could it do to help medical professionals operate maybe more, what could it do for them to help them navigate the complexities of their job? I don't know. I think it's such a new space that I'm like, I've got thoughts. Then I don't have thoughts, then I have perspectives and I'm like, all these perspectives are valid and important. And I'm like, yeah, I'm trying to figure it out.

David Elikwu: The last thing I might ask is, it might be a follow on from what you were just mentioning how we could, for example, use AI to support healthcare professionals. I'm interested in your perspective on How do [01:39:00] jobs of the future interface with ai, and we're trying to use AI right now to enhance our creativity, enhance efficiency, enhance productivity, a bunch of these things supposedly. Where do you think the room will still be left for humans to do what they do best or what they have done best? Because there is a large proportion of these things. Where you might not need to work in the same way that you have previously, you might not need to do all the precursor steps in the same way that you have previously.

For example, with writing, you can type something into Chat GPT, and it is faster than searching Google and there are things that I would've spent, you know, half an hour trying to search Google and trying to make sure that I've got the correct answer. If I was writing something and I wanted a historical reference, I could search and I could go through 10 different websites and I could, or I could just ask Chat GPT and it would just gimme the answer. It would gimme some additional context. So how do you think the future workforce will interact with AI? Do you think it [01:40:00] will simply stay as an aid, or do you think there's an extent to which it might replace or change forever some of the roles that we currently know.

Alex Fefegha: I think our first Chat GPT needs to be good at giving true academic references. Cuz I literally saw the other day where someone close to me was writing, I think they were writing an essay and they used like an indicational AI tool to help them and then it gave them a citation. And this citation pop up were trying to Google on the internet, the citation like, it looks so legible. Like if I was marking an essay, like in the past I've marked essays and, you know, when I've marked essays in the past, it might be certain citations to student who's used where I'm like, oh, this sounds proper interesting. I should actually, you know, let me read into more. I was pop up digging the internet. The conclusion in the end was like, yeah, this don't exist, man. This is some

coherent text and it sounds real, but it's questionable. So at this period of time you need [01:41:00] to verify if what it's saying is actually truthful or it's not just coherently sound and text. And you have seen that in current, sort of cases and stuff like that. I think for me right now, I am constantly on this journey of understanding like AI's role and society. For me, where I'm looking at a lot is mostly healthcare because outside of the artistic side of doing AI and generative AI, you know, Comuzi's focused a lot for the last couple years since Covid has been about healthcare, bringing emerging technologies to healthcare and how we can help improve, develop new access, new models of care for different communities across the world and stuff, that's like a big area for us. And so one of the areas I'm really interested in is the role of AI and healthcare so, one of the things we'll be focused a lot more in 2023. It's actually that route. Like, does AI have a place in healthcare right now? what way [01:42:00] does it have? And then the second part of the project is looking in that, examining that maybe 10 years from now, what are the different products and services that will be powered by AI that will sort of support medical professionals to do what they need to do?

And I think, the world is changing ever so fast and we're constantly having to figure out where this new world we're in is all the time, but the world is moving way too fast in my opinion. I think It should slow down cause the world is burning and a lot of these things and we need to fix the structural issues we have. But you know, our desire for the new and the shiny keeps us running like this. So right now, I really don't really have a clear answer. I think there is a potential of it being a very good aid. I think that's how we should see it. Not as this magical tool, but as a very well sophisticated aid like Jarvis and Iron Man and stuff. I think there's a [01:43:00] lot of potential for it to be a Jarvis. but I think, yeah, it's such a new space that was a growing space of conversation and contention that I don't have a one set head space on it.

David Elikwu: I hear that. But anyway, Alex, thanks so much for, taking the time. I really appreciate it and it's been incredibly interesting to hear lot of your different views and perspectives, particularly coming from someone that has been in this field and has seen a lot of these tools progress over time. 'Cause I think that is the big distinction a lot of the rest of us as are laymen, is that, you know, you just see these tools pop up and it seems like things pop up overnight it's been interesting to see. I mean, even from the nuggets that I've been getting from you over the last few years, seeing how some of these things have changed and developed and iterated over time. So I think it'll be really interesting to see where all of these things go in the future as well.

Alex Fefegha: Yeah, I'm proper intrigued, interested. It's a space of exploration. It is an area that I'm gonna double down on and try to understand more and more. But like I [01:44:00] said, computers are stupid. So I'm gonna make sure that wherever, for me, in my opinion, wherever I put my hands to AI, it's potential for failure isn't gonna harm someone's life or anything like that. And it's in a place where it's a bit more controlled and stuff. Cause I think that's something we have to bear our mind as well. So, AI is already embedded in almost everything we use every day, it is changing the world. But to go in this whole dramatic full scale change, I don't think we're ready for that. I don't think the technology's ready for that, and I don't think we should even do it. But that's a conversation for another day. But, now thank you so much for having me on the podcast.

David Elikwu: We love it, man. Thanks a lot I appreciate it.

Thank you so much for tuning in. Please do stay tuned for more. Don't forget to rate, review and subscribe. It really helps the podcast and follow me on Twitter feel free to shoot me any thoughts. See you next time.[01:45:00]

Share this post