[00:00:00] Peyton: Hi everyone. And welcome back to our CareerCon all about AI. Um, I'm Peyton, the designer and developer, um, here at Hundo. And today I'm joined by Elaf. Hi Elaf. Um, could you just sort of introduce a little bit about yourself and what you do?

[00:00:26] Elaf Deyab: Yeah, of course. Uh, great to meet you. Um, I'm working on Caspa AI.

It's an AI agent that helps e commerce sellers create and optimise product images to increase their sales. Uh, before that I studied engineering at university. I did a dissertation in deep learning, which was my first insight into AI. Um, and then left to work in investment banking for a couple of years.

I worked at Goldman Sachs, um, before leaving to start another company. I've worked on a bunch of different projects, uh, throughout the most relevant being. A digital marketing agency where we helped brands create product photos and, and ad campaigns online.

[00:01:00] Peyton: Lovely. And we're going to jump into a few AI sort of related questions here.

Uh, just to kick us off, um, in what ways is AI revolutionizing the e commerce space and how does Caspa AI specifically assist companies in integrating, um, AI with human workers to foster that innovation and enhance sort of e commerce productivity?

[00:01:22] Elaf Deyab: Yeah, there are loads of ways that AI has kind of started to help in the e commerce space.

Um, things like personalisation or automating customer service, uh, optimising their inventory management and also helping with like pricing and in loads of different ways. With Caspa specifically, um, we're helping with photo optimisation and photo creation. So, um, e commerce brands lose up to 75 percent of sales due to poor imagery and We know that AI can change this.

So we basically integrate AI with human creativity, allowing people to create lifestyle photos with realistic human models. And we can, they can also AB test different ethnicities in order to reach different demographics and kind of have a broader target market. And then ultimately this will all lead to an increase in sales while saving them time and money.

So they don't need to organize photo shoots anymore, higher models. Facebook, real estate, do all of that stuff. Um, instead they can just very quickly, rapidly generate the images that they need and post them online and focus on whatever they need to focus on next.

[00:02:19] Peyton: Fantastic. And yeah, A B testing was something that initially came to my mind.

I think that continual testing is a very important. Skill to have, um, particularly going to any industry, really, but, um, quite applicable in the AI space to just expand on those skills a little bit. What other skills could young people need to sort of develop to thrive in an AI driven future? And how can they begin to start building those skills now so they can stay ahead of the curve?

[00:02:46] Elaf Deyab: Yeah, I would say the first thing is just getting stuck in, um, not overthinking, uh, uh, kind of learning about AI or starting to kind of build a game or create something. Um, and also not relying on, on people to handhold you. I'd say, um, there's loads of like hackathons hosted by colleges and universities, uh, If someone's interested and they don't know where to start, those are like great places to go and, and meet people interested in building the same thing.

And also just get, you know, hands on experience working on something, but also there are so many YouTube videos and tutorials online that are free. And you, you could literally just kind of duplicate what people are telling you to do. There's like specific practical skills to learn, like learning how to code or learning stuff about a specific niche, but everything ultimately relates to that.

To problem solving. So if you learn to get better at problem solving, learn to like problem solving, uh, then that will just put you miles ahead and it compounds over time. Um, if you commit to doing it now, even if it's like 10, 15 minutes, an hour a day, um, in five years, you'll be miles ahead of everyone else.

[00:03:47] Peyton: Yeah, totally agree. And I think, um, having conversations with, with people around you and being naturally quite curious, um, is an incredibly valuable trait to have. Um, it's even. Sort of going beyond AI at some point in the future, there will be new technologies and new things that happen and remaining curious about whatever it is that is new and cutting edge.

I would personally consider it quite important. Um, and then also sort of, um, Looking at mentors in young people's lives, parents, educators in preparing those young people for a future of AI. Could you give some advice for parents and educators that are looking to support young people that are looking to get into sort of content creation in the e commerce space?

[00:04:34] Elaf Deyab: Yeah, of course, I'd say, uh, encouraging curiosity about technology. Uh, I think there's a lot of the time whenever there's something new that people aren't sure of, um, instant reaction is to be scared, but the fact is that it's just getting better over time. And so if they're able to, to, to encourage learning about it, learning about how it works, learning about how things are built, um, just digital literacy in general, it's super important.

Um, it's the future, so you may as well understand it. Um, and so for a parent or a mentor or an educator, if they can provide, Access to coding and AI courses, hands on projects, just like environments that, that kind of fosters this, this, um, like willingness to learn about, uh, AI and tech in general, then that will, that will do way more than, than they can imagine.

[00:05:20] Peyton: Yeah, absolutely. Um, and then another sort of very important and topical consideration at the moment, um, surrounds. ethics and morals of AI models. Um, how can AI tools like Caspa AI ensure responsible and ethical use in content creation and what are those potential risks and how can we start to look at addressing those and solving those ethical issues?

[00:05:43] Elaf Deyab: Yeah, this is like an ongoing conversation. And I mean, AI is fairly, it's a fairly new innovation. Um, regulators are still learning how to deal with it. Uh, and, and kind of a larger scale, I think there's like a bunch of risks that include kind of stereotypes or perpetuating stereotypes, um, infringing on IP, um, also just like generally misusing, uh, generated content.

Um, so in order to address them, we just kind of constantly have to be ahead and trying to understand how people could misuse it. Um, there needs to be like clear ethical standards. And I think that's something that's kind of top of mind on kind of for people who are working in this space. And also just like user education, just educating people about the space, educating them about how they can use it and what is, you know, You know, the direction of things that they should or shouldn't do.

Um, in terms of Caspa, uh, we can encourage responsible and ethical use in content creation generally by kind of, uh, implementing guidelines of, of kind of use cases and, and, and kind of focusing on data privacy. We're avoiding biased outputs based on the training of the models that we're working with. Um, and generally trying to be as transparent as possible with our processes and, and other AI processes.

I think transparency is probably going to be. Something that's super important for anybody who's building in this space, um, to help people who, who might be scared of it.

[00:07:02] Peyton: Yeah, absolutely. I think, um, education, teaching people how to use these tools responsibly, and then also making sure there are appropriate guardrails in the actual tools themselves.

I saw a very interesting new development from Anthropic recently, some research from them, where they were able to train a secondary AI model to identify monosemantic meanings from their data. Traditional large language model, um, which they were then able to, um, sort of dial up and dial down, um, specific traits of their models.

Um, and some of those, some of those could be anything. There was a, a sort of, uh, an AI space meme, if you will, about them dialing up the Golden Gate Bridge trait, and then the AI model thought it was a Golden Gate Bridge, but, um, very important research in then being able to dial up Traits that are related to safety and ethical output, um, and being able to further fine tune those and doing that in a way that isn't just pre prompting, but also at a more fundamental level so that it can't be as easily jailbroken.

Um, and then. Leading on a little bit, um, in your view, how do you envision AI reshaping the workforce in the next five to 10 to 20 years? Um, and what proactive steps could educators and companies take now to prepare for those future changes?

[00:08:23] Elaf Deyab: Yeah, I think the biggest thing that we're going to see is, um, just Automating manual tasks.

Uh, that's something that, you know, uh, AI is pretty good at, and it's just going to continue doing that over time. Um, also just decision making there's some things that are kind of pretty standard and AI can help with, with just like deciding on certain decisions. Not all of them, obviously, there's obviously going to be a lot of human input involved.

Um, but, but because it's a new innovation that there's also going to be a lot of new jobs that are created, things that we might not even be able to imagine. Now in 10 years, there'll probably be jobs for it. Um, I'd say to educators, um, focusing on, on kind of AI literacy and just encouraging it and, and for companies, uh, investing in upskilling people, um, making sure there's a culture of continuous learning.

Um, and then also kind of like we touched on before the, the ethic, um, ethical and transparent kind of processes, just making sure that that's kind of top of mind for all the companies that are in the space.

[00:09:19] Peyton: Absolutely. And I can definitely see a world where AI, it may replace some jobs and some tasks, but I can, I can definitely see an increasing world where it's more augmenting a traditional human role with increasing sort of technological use.

Um, I've definitely experienced that in everyday tasks that I use. Um, it. Doesn't replace any task in, in its whole, but it, it definitely supplements it and helps it and, um, sort of does increase, um, productivity. And there is also a sort of certain learning curve there to being able to know which tasks a AI might be good at, um, like, uh, when it might be good to consult something like ChatGPT, and then other times when it might be best to use another tool, a traditional, um, tech tool, um, and obviously different tools have their traditional pros and cons.

Um, and I have one or two sort of in one final question, questions for you, um, today, Elaf. Um, what excites you and what scares you the most about the future of AI?

[00:10:27] Elaf Deyab: That's a good question. Excites me. I think, um, when we look back at all the other kind of tech innovations that have happened, things like planes or things like mobile phones, there's obviously certain negatives, um, that could be associated and things like crushes and things like people may be on their phones too much nowadays.

But I think the benefit has been, it just far outweighs the negatives and I'm excited to see what that's going to be for AI. Just saving people time and just bringing people closer together, hopefully in the future. And I think the, the, the risks are mainly ones that stem from humans, like kind of teaching the AI the wrong things.

So I think as long as we kind of are focused on like the ethical use of this kind of tech, and that's something that's like super important. And then hopefully people won't misuse the tech train on, on kind of models that, that we don't want it to be. Um, Uh, kind of teaching people or whatever in the future.

[00:11:21] Peyton: Yeah, absolutely. I suppose another risk that comes to mind is wanting to make sure that Um, everyone gets an equal opportunity to be able to use these these tools, um around the world We certainly don't want to leave people left behind. Um, and wanting to make sure that um, everyone gets a chance at developing them as well.

Um, you definitely don't want to uh, sort of Guard or or garden wall in a certain new technology into a certain specific set of, um, companies or people that are able to use them. Um, I definitely personally strongly believe in a sort of open framework, um, as much as possible. Well, thank you very much, Elaf, for our conversation today.

I really enjoyed talking about it and hopefully a very exciting future of AI. If people would like to connect with you and learn a little bit more about Caspa AI, where is best to connect with you in and around the internet?

[00:12:21] Elaf Deyab: Yeah. Uh, so on LinkedIn, on email, uh, elaf@caspa.Ai and our website is www.caspa.Ai as well. Um, so feel free to reach out and thanks so much. It was great chatting to you too.

[00:12:33] Peyton: Thank you very much indeed. And thank you to everyone else watching, um, have a lovely remainder of your day. Take care.