1Mby1M Virtual Accelerator AI Investor Forum: With Rajeev Madhavan, Clear Ventures (Part 4)

Sramana Mitra: Now, are you seeing the AI startups also starting to do services work? I have seen several instances where they have AI capability, but they’re actually going to market with large projects where they provide product and also do the services as one solution.

Rajeev Madhavan: It is a good way to enter the market. For example, in the tax company that I mentioned, they have a lot of people who are providing services for the entire small, medium businesses to do taxes. It’s not necessarily the long-term plan, but it gave us the ability to get going. We actually acquired a couple of companies that brought in the technology and the relationships with the customers and added to it our technology to improve the margins and productivity. Now we can go and sell to service providers and to everybody and show what it means to use our tool. It’s better than waiting for service providers to help you.

By the way, this is nothing new. When I did Magma, as you mentioned around 2000, I think we had a service arm which did couple of very large chips of that time. I can’t wait for somebody else to get going. I just had to get a bunch of services and get going. Creating the tool in that context to understand things may not be my long-term plan. I’m doing it as a means to create my tool, get the flow captured, and get my tool out in the market.

Sramana Mitra: Well, in the analogy that you just gave of chip design, the expertise is finite, right? It’s not like there’ve ever been millions of high-end chip designers who could do whatever on the fly. Right now, AI is in that state. High-end AI expertise is not as ubiquitous. Inside Fortune 500 companies, this expertise is not randomly available. So, as that expertise develops and expands, there will be more possibilities. But also in the meantime, the startups are getting to learn by actually solving the problems. They’re also learning about what other functionalities you need to bring into the products. It’s a win-win across the board.

Rajeev Madhavan: It basically helps understand what it takes to get your great idea into market, right? At all costs, get in and get the features done. Just make sure that when you do that, it’s the tool that you’re selling and not the expertise as a card service.

Ten or fifteen years from now, we’re going to have AI as knowledge for most people. All of these models getting served improves, and we can operate. To me, the change of what is going to happen is the computer science experts. It will move a few experts to everybody being able to contribute to a certain degree. That’s a very good thing to happen, but it will take 10-15 years. You as a startup don’t have that 10-15 years to get the ball moving. You’ve got a couple of years to be in the lead and establishing your footprint by the time the entire market community is behind this new transition.

Sramana Mitra: Well, we are in a very fast-paced innovation cycle mode right now. In the rest of this decade and the next decade, things are going to move really fast. So I’ll actually end this segment with my last question.

How are you viewing the evolution of general AI and superintelligence? And how does that inform your investment thesis in vertical AI? Is superintelligence going to make vertical AI unnecessary? How do you parse this question?

Rajeev Madhavan: I think things like AGI, super intelligence, and words like that are dreams of the AI community. As to things that will work, I don’t think we are even near that. At the end of the system, people have to just look at it and say if the patterns of things have been seen before and if we are going to reach a point where the system can do a repeat.

But having to say an AGI of the model and making cognitive decisions beyond what is today obvious in the data and the model itself, it’s going to take a lot of time. The transition to the current state is going to take 10-15 years. There could be an AGI in some specific localized area where every aspect of it can be trained. So much money has gone into the training to do that. Much of the systems can be engineered such that the inference of the data needed to make adjustments can be done, but I don’t think everything in life is that easy to be done. That in my view is for the next generation to carry on and do the work. I think it’s a great pioneering work that needs to be done. If this one step is carried through well by startups, they could end up in great positions over the next few years.

Sramana Mitra: You know, my take on general AI is that it doesn’t really matter because there is at the moment, the opportunity in front of us in vertical general AI, let’s say, the workflows, the APIs, all the integration and kind of real automation in the context of a particular problem domain is so big. And so the productivity that it can achieve is so huge that’s at the moment, at least, that is really the opportunity.

Rajeev Madhavan: Yes, akin to making every human a superhuman in that specific domain or whatever it is. That is bound to happen. We’re all going to become able to get summarized knowledge of what happens. Like for this podcast, somebody could generate a three-line summary out of a system.

That’s great. I don’t know what it’ll look like, but it certainly is great that it’s going to get down to that level. As you know, over the period of this transformation to smartphones, humanity has lost the ability to read long things. They need it in snippets that are shorter and shorter. All the summarization at a higher level and summarization of the core content has taken a toll on our social relationships, but clearly it’s going to change the way we have been operating to be very quick and get to the data and arrive at solutions very quickly.

Sramana Mitra: Well, the fact that we cannot do difficult things is not necessarily a good thing. In our evolution, we’re losing capacity for rigor. This is not a good thing.

Rajeev Madhavan: No, I agree with you completely on that. It’s the same things that have happened with smartphones. In early 2000s, when smartphones came in, we didn’t anticipate the impact. Now there’s a lot more impact in universities where people are not mingling as much with each other. The suicide rates are high.

Sramana Mitra: All skills have gone down the drain. That’s a very bad thing.

Rajeev Madhavan: And we’re going to have similar impact here because our ability to learn is going to get reduced further and further in this business.

Sramana Mitra: Well, yes, but that is a longer conversation for some other time. This was a fascinating and very topical conversation. We will continue the conversation in a few months. Thank you for coming and sharing your perspective, Rajeev.

This segment is part 4 in the series : 1Mby1M Virtual Accelerator AI Investor Forum: With Rajeev Madhavan, Clear Ventures
1 2 3 4