Loading summary
Transcript30 lines
- [00:01]
Audible Promotion
Audible's best of 2024 picks are here. Discover the year's top audiobooks, podcasts and originals in all your favorite genres, from memoirs and sci fi to mysteries and thrillers. Audible's curated list in every category is the best way to hear 2024's best in audio entertainment, like a stunning new full cast production of George Orwell's 1984 heartfelt memoirs like Supreme Court Justice Ketanji Brown Jackson's lovely one. The year's best fiction, like the Women by Kristin Hannah and Percival Everett's brilliantly subversive James Audible. There's more to imagine when you listen. Go to audible.com imagine and discover all the years best waiting for you.
- [00:49]
Instacart Promotion
Why get all your holiday decorations delivered through Instacart? Because maybe you only bought two wreaths but have 12 windows. Or maybe your toddler got very eager with the Advent calendar. Or maybe the inflatable snowman didn't make it through the snowst. Or maybe the twinkle lights aren't twinkling. Whatever the reason, this season Instacart's here for hosts and their whole holiday haul. Get decorations from the Home Depot, CVS and more through Instacart and enjoy free delivery on your first three orders. Service fees and terms apply at the moment.
- [01:20]
Hamayan Shaikh
You know you don't get to see startups come to the market. You don't capitalize on the ability to invest in these AI startups. But that's going to change because you could own your agents. You can actually own the AIs that sits behind the agents. You can even own foundational models which people have built and they could look to raise funding.
- [01:48]
Jen Senasi
We all want to believe in something.
- [01:51]
Hamayan Shaikh
In Chinese, belief also means trust. We want to trust our systems and each other so we can build a future beyond what we know. Join us at Consensus Hong Kong, where belief becomes real.
- [02:04]
Jen Senasi
Hello and welcome to Markets Daily, hosted by me, Jen Senasi. On this show, we navigate the currents shaping the crypto markets, providing insights against the broader financial landscape. So whether you're actively trading or simply fascinated by the volatility that is the crypto markets, this show is your compass to understanding what's happened, where we are, and where we're going. On today's show, we're joined by CEO of Fetch AI and the Chairman of the ASI Alliance, Hamayan Shaikh. Hamayan, welcome.
- [02:30]
Hamayan Shaikh
Thank you. Nice to be here, Jen. Appreciate it.
- [02:34]
Jen Senasi
Of course, as you know, on this show we talk about all of the trends that shape the crypto markets and AI is a really, really strong Trend. It has been one throughout the year, but it's really ramping up as we head into 2025. Talk to me a little bit about how you see AI shaping the future of crypto trading.
- [02:51]
Hamayan Shaikh
Yes, AI is just a general trend anyway, but I think it's really well suited for crypto because of several reasons. Decentralization effect of people owning their own data and trying to be more in control of machine learning, all the AIs that they create. So I think it's just about the right trend and I think the technology is mature enough to come into the web three space and the crypto space. And another interesting part here is the digital currency itself with AI to AI trading, the microtransactions, it fits beautifully. So I think on that front that's pretty good. The other really interesting thing is the kind of the business dynamic of AI is going to change the way the paradigm shift will change. So there will be a lot of people owning these machine learning and AI models. And those models are actually worth quite a lot of money. So effectively you're kind of going to see when you talk about trading, you're going to see AI models actually overlapping with cryptocurrencies and becoming kind of a thing of its own. So we might see in the very near future some really interesting combinations of AI and crypto trading.
- [04:15]
Jen Senasi
Tell me a little bit more about that. Give me a practical example for folks who are really trying to make sense of what the future for AI and crypto trading could look like. Talk to me about what you just described might look like for a trader.
- [04:27]
Hamayan Shaikh
Sure, there are a couple of, couple of things we can, we can talk about. So one is, so fetch AI which launched like five years ago, we've been working on agent AI agents and agent based infrastructure. So we launched various AI agents which people can own, people can actually deploy and they can do trading, peer to peer trading for you. So if you look at where we've come from on Defi, you have this central contract, yes, it's decentralized and people provide liquidity. But there is a central point of failure in the middle where you interact with that contract and you deposit all your liquidity. A different way of looking at it in the AI world is now agentic system where you actually just hand over not your keys, just your assets to an agent and the agent then goes and looks for other agents to transact with. So a true peer to peer, decentralized in your control. So what does it actually do? It stops the rug, pulls it stop market manipulation. Your agent is in your control, so you're not relying on a central point of failure. So I think that's quite an interesting defi angle. The other interesting thing which we're going to see is in decide decide is really going to become quite important because the science and research and the funding, the way the funding work with universities is going to change. Research will accelerate quite high because of the generative AI, the machine learning models and the compute that's becoming available. So people in crypto will get the opportunity because of the way it's structured, we get the opportunity to actually own a part of it. Because at the moment you don't get to see startups come to the market. You don't capitalize on the ability to invest in these startups or AI startups. But that's going to change because you could own your agents, you can actually own the AIs that sits behind the agents. You can even own foundational models which, which people have kind of built and they, they could look to raise funding. So we might see another new funding model which is going to kind of come to the market which, which is kind of not ICOs or IEOs or IDOs, but more real assets, real values, which sits in AI being owned by crypto kind of traders.
- [06:55]
Jen Senasi
That's really interesting. You know, the crypto folks are real early adopters. So if we zoom out a little bit and we, we look at AI in general, there's this real fear when it comes to AI and AI agents. I don't know if that fear really exists amongst the crypto folks who are experimenting with it. But you talk about giving your assets to an AI agent that can then do peer to peer trading with other AI agents. Is there any drawback or challenge in giving AI agents that autonomy over your assets?
- [07:26]
Hamayan Shaikh
I think it will be quite progressive. You will not be able to just say, okay, well here's, here's a $10,000 and go and do whatever you like. What would happen is that you will, you will say, okay, here's $10,000, I want to invest or buy or trade X, Y and Z tokens. And then here's my kind of price. I mean, just like in a normal exchange you'll have stop losses and you will be able to kind of say to your agents, do that. But ultimately when the agent is about to transact, you have to still authorize it. I mean, we have a demonstration of that in an exchange which is a peer to peer agent based exchange which has come out of Fetch, called Metalex and you can actually do, do these transactions right now. You can see you have agents, they act like your agents and they're your agent. They trade on your behalf. So they can't just go and do whatever you want them to do. They will listen and they will come back to you for approval. And you have to accept the transaction before anything happens. Now that will kind of change perhaps, because as the security, the safety of these agents become more, it becomes tighter, it becomes more stringent, it's more controlled, then you perhaps will allow them to do more autonomous trading transactions.
- [08:55]
Jen Senasi
We've talked a little bit about what's going on at Fetch, AI and AI agents. And we'll come back to that later on in the show. But I want to touch on some recent news. The ASI alliance unveiled ASI Train. Talk to me about what that is and how you plan to see it develop or your roadmap for 2025.
- [09:12]
Hamayan Shaikh
Yeah. So for those of your listeners who don't know, ASI alliance is a merger of four projects now, which includes Fetch, AI, Singularity, Ocean and Kudos. And they cover different slices of the tech stack. So we have compute, we have data, we have algorithms, we have agents. So the whole thing covers right from the application layer, which is the agent based, down to the compute, which is pure GPU compute power. So that's the ASI Alliance. But we are launching an initiative called ASI Train. Now, if you look at what's happening in the market right now is there was a lot of hype over large language models and OpenAI and several others built those and improved those, and you can see the advantages, the benefits of it. But they are still large language models. They need to be customized or new models need to be built, which are called foundational models for different verticals. So although they understand the language, they don't understand the verticals. So for example, if you want to create a specialist agents, which could be part of material science, which would be doing materials discovery, molecule discovery, organic, inorganic, you're finding a new battery material or trying to find a superconductor, you need to build, you need to use all those machine learning models. You need to build these foundational models which sit within a vertical because LLMs are wide and they're language models. You need specific models to do and accelerate that research. And, you know, 100x the discovery speed at the moment. And that's the new next trend. And I think it will become more and more popular because you still have these universities now which are struggling because of the commercialization of Technologies and students and researchers kind of need the ability to launch their machine learning model. They might not be business people, but they might be machine learning people. So what you could do is you could train these foundational models and these foundational models can be trained in a decentralized way where all the crypto folks can actually participate in it. And that's what the ASI trained initiative is. And then the revenue will come from people who take the inferences from these models. So if I say, hey, I have a molecule, I want to see what other molecules can interact with it and how you can submit that to this foundational model and you can get the output. Now the Training might take $10 million, but you don't have $10 million. So you go to the Web3 space, people can train it and for inference might cost you only $1,000, but that inference is very valuable. So ultimately these foundation models become very valuable. So they, you know, I mean, if you think about, they won't go like meme coins, because meme coins can be crazy. I mean, we're talking about traders. But what you have here is something which could be quite valuable. Very quickly as you train and refine.
- [12:23]
Jen Senasi
Them all, dig a little bit deeper into that economic potential. For me, what might that look like? I know I keep talking about the next 12 to 18 months, but what does that look like? How does it unfold?
- [12:35]
Hamayan Shaikh
So let's say you have a physics model or material science or drug discovery model that somebody is building, like an, either an open source or university students or a group of researchers. So what they can do is they can deploy it on the Web3 space. And actually this, quite interestingly, this is a beautifully Perfect model for Web3 and crypto space. Because we have been struggling with getting true utility from different verticals. We quite insular in just token kind of based trading. So this brings that new facet into crypto trading where you can actually build this model and you can deploy it. And the, the people who actually invest in it as a model, they get a stake of the model. So for example, let's say day one, the model is $10 million or $1 million or 100 million, whatever the value is. Let's start from a million say. And this is without not much training. So then what you do is you open up this, either a dao or a pool where people can put money in and you own a portion of that, the model. So just like, you know, if you have a token project, you have 100 tokens and people can come in and buy tokens. So they own that part, they become the community, they interact with the model. So now as you train this model and other people train the model, which is, which is actually every single time you train it and fine tune it, the value starts going up. So as the value grows, your share of the model actually grows as well. So you own that value increase. So that's how the economics could work.
- [14:22]
Jen Senasi
I touched on this a little bit at the beginning of our discussion here. But if we go back to AI agents, we stay in the context of these foundational AI models. Are there any ethical concerns that arise that you're discussing as you continue to think about these?
- [14:38]
Hamayan Shaikh
Of course. I mean, this is a new field and there is, there is quite a lot to discuss. I mean, who owns the ip, where does it sit, who uses it, for what purpose? That's also quite interesting. But I think, I think, I generally feel that we need to let the technology evolve to see what the use cases are before we start regulating these things. So ethical concerns for me would be, I mean, actually it actually improves the ethical concerns because if you think about building these foundational machine learning AI models, you know, it's quite open source, it's out there in the market, you are giving opportunity for everybody to participate in it and everything happens on chain. So you have auditability, who transacted with it, how did it get trained, who trained it. So all those things actually bring that barrier down and it enables you to audit what you've done and how the AI is going to behave. So on that front, I think there's, there's a positive side there. But I think you pointed out on agent side, you know, how do you make them autonomous? What if something goes wrong, who becomes responsible? So as I said, I think if we, if we actually deliver the products in a more controlled and auditable fashion, we can overcome that kind of hurdle quite easily.
- [16:03]
Jen Senasi
And I want to, I want to talk a little bit about AI agents versus like human intuition. This is something people continue to talk about not only in the crypto space, but everywhere that AI is applied. How, talk to me about that balance. How do you think about that balance? You know, do you think AI agents eventually will dominate crypto trading, or is there still going to be a need for human intuition and strategy and the two will really have to kind of work together for us to really see the benefits of AI and AI agents.
- [16:36]
Hamayan Shaikh
I think we are starting to get to a point where AI agents will be quite good. So I mean, one of the reasons we called this alliance, Artificial Superintelligence alliance, is because these AIs and these agents will become super intelligent. Now they, they become super intelligent. And the definition of super intelligence, or AGI or whatever intelligence is, it varies from different people. Some people think it's going to take over the world and it's going to be so clever. I have a kind of a bit more conservative view on that. I think we will always have humans in the loop and I think we need to kind of go into that direction so that we accept that, that there will be like automation tasks, there will be specialist information which will come from these agents. There's no way a human, you know, it's like all these foundational AI models that we're going to be building. A human isn't going to kind of run them in their head. But what they can do is they can take the inference from those models and utilize it. For example, sentiment analysis combined with, you know, ta graphical analysis of a token or a currency and then combining the two, plus any news that's coming in. Humans can sometimes just act more on intuition, which is pretty good, but you will have the tools to actually make that intuition even better. So I am more of the opinion that you'll have a human in the loop. However, the intensity of that human's research will come down because the research will become very available and very coherent and very understandable. So they can make that decision a lot quicker.
- [18:18]
Jen Senasi
Yeah, I really agree with you. I always think of the evolution of AI as the evolution of humankind. The two will work hand in hand. Like it's not going to. AI is not going to take us over. I mean, maybe one day I'm going to be eating my words. I can't predict the future, but I really think that it's an extension of us and just the evolution of us.
- [18:36]
Hamayan Shaikh
There has to be a step change, I think, to come before we get to that point.
- [18:40]
Jen Senasi
Yeah, yeah, yeah. All right, as we wrap up the show here, if there's one piece of advice that you could give crypto traders looking to leverage AI today, what would it be?
- [18:50]
Hamayan Shaikh
I think if you're looking for real value, real world value, AI is here to stay. Has been for the last 20 years. So we've been using it and I think the acceleration that is happening right now is going to continue. So if you are looking to come into trading, I would highly recommend as part of your portfolio, having some AI assets in your portfolio which you feel capture the value which is about to be unlocked.
- [19:20]
Jen Senasi
Any particular assets or projects that you are watching closely that are really interesting to you?
- [19:27]
Hamayan Shaikh
Sure. I mean we do. I watch Near Protocol, which is quite interesting project. I mean, I'm not going to name the alliance projects because there's no point, as I said. But we also look at Bittensor. It's a good project. Render is a good project. I think these are pretty decent projects, but I think the space we all exist in is quite slightly different. So they're not like, I mean, again, also ICP is another good project. All these projects exist in their different spaces and the different niches and the market is actually quite considerably big. So, you know, I think having a few of these in your portfolio will probably be an interest. But this is not a financial advice. I guess I'm just saying it as I would see it. Yeah.
- [20:17]
Jen Senasi
Mayan, thank you so much for joining Markets Daily today. It was a fascinating conversation.
- [20:22]
Hamayan Shaikh
Appreciate it. Thank you.
- [20:24]
Jen Senasi
And thank you to our audience for watching. If you enjoy Markets Daily, please head on over to the Coindesk Podcast network that is available on all podcast platforms and hit subscribe. If you prefer to watch. We are on YouTube. Hit subscribe. Give us a thumbs up. Thanks so much for coming on this journey with me to better understand the crypto markets and we will see you tomorrow.