The Exchanges

Q Yeah, no, that's a fantastic takeaways to have, and, and kind of, especially going into now, your current role with Glasswing, and I do want to start with the investment strategy there, and it's focused on AI, and as you told me before, more specifically, narrow AI. So, 10 years ago, this wouldn't really have happened. So, Why is AI firm focusable theme now today? What's your take on this?

A To your very last part of your comment and question, AI and machine learning and even deep learning technologies have existed for decades, to be perfectly honest. What has transformed or what has changed the market timing for adoption is this emergence of what we at Glasswing call pervasive connectivity and And, um, the inputs and outputs to it. So by pervasive connectivity, we're really referring to the notion that consumers and enterprises now have many, many, many points of connection. We don't see convergence to one device. We see the existence of many devices around us, wherever we are in the home, in the car, um, in our offices, educational institutions, wherever the case may be. Where we're constantly connected and constantly on through different devices and means. That pervasive connectivity has enabled the emergence of a new wave of multitudes and terabytes of data that now can be leveraged to train models on the AI side, can be leveraged to essentially bring sort of the AI promise with machine learning, but also the other facets from speech recognition to face recognition to social cognition, et cetera, To reality. Couple that with the fact that mobility as a revolution, if you will, has happened, and storage costs with cloud computing have come down dramatically, make for the right market timing for narrow or applied AI to emerge.

AI assessment note: “What has transformed or what has changed the market timing for adoption”

Q Taking a slightly meta view and looking at just the kind of disruption cycles themselves now, where would you say we are in these cycles and curves?

A In our view, again, and, um, we fundamentally believe that we were entering the third disruption cycle, so, and we think of the first as it being the web in the, you know, in the, you know, in the, Sir Timothy Berners-Lee, out of your home country, um, invents the web, so we see the disruption that emerges out of the digital wave. Then, um, we see the social mobile wave, and that disruption, or that wave, in our view, the revolution is complete. There will be, there is evolutionary or incremental, um, you know, improvements in nature, and we'll see new products, but from a transformational point of view, I think that, um, wave is complete. And now, with this notion of pervasive connectivity, where you have the right data, and the right, um, you know, technological drivers and inputs, We think that AI, the AI wave has already, in fact, begun. We're probably three, four years into it, and we expect that, as in prior waves, the adoption cycle for AI-powered products and platforms will be shorter. From the web to the social mobile, we saw adoption half from 14 years to seven years, and we expect, um, you know, a similar rate, if you will, in the speed of adoption with the AI wave.

AI assessment note: “we fundamentally believe that we were entering the third disruption cycle”

Q What, why is that increased speed of adoption? Is it the storage? Is it the data? What makes it so much shorter?

A Yes, and yes, and yes. I think it's that, I think we, it's the, um, millennials and the younger, um, generations that, um, how tech savvy they are, and how quickly that adoption is taking place, and they're, then pushing to sort of the older generations. I mean, if you think about it, you're young, but But you, I don't think, remember a world without web and without digital presence, and today's teenagers don't know a world without mobile, and it's part of their nature, so they are very much driven the mobile social wave, and I have a three-year-old, and I think she's going to be an AI native with not just sort of the interactions and a very redefined relationship with technology, but also The whole world of social robotics and the increased role that, um, robotics will have in our lives, which at the core really has, um, applied AI and even one day possibly general AI, 2030 years from now.

AI assessment note: “it's the, um, millennials and the younger, um, generations that, um, how tech savvy”

Q Biggest mentor to you and how it came about?

A So it's interesting. Mentors have changed over time for me. Um, they've always been people that have been adjacent to my industry, but not quite in the industry. So they have been guiding me without being too much into weeds, and I've always been grateful for that. But they've changed over time as my career has advanced and altered over time. So I would say the steady figure in my life has really been my mother. And both from a mentoring and role model point of view, um, she ran a 3600 person enterprise, and I've seen the, what it takes to lead, and, but also she raised myself and my sister to be pretty strong and independent women, and sort of, I hope to have the same chance to instill those values in my family.

AI assessment note: “I would say the steady figure in my life has really been my mother.”

Q I'd love to hear, you said about your time at Microsoft there. What were some big takeaways from working in such a, such a behemoth of an organization?

A Well, behemoths and dinosaurs die slowly, so the ability for a large company to disrupt oneself is quite difficult. If they succeed in doing it, it's usually the exception, and I'm often hard-pressed to find such exceptions. Perhaps Apple, because it nearly died and then Jobs came back and kind of reinvented it and resuscitated it. But, um, so I think driving change within a large, um, tech company is quite difficult because you have to essentially, um, be able to cannibalize your own current revenue streams with, um, public market pressures. I also think that from my perspective, when you're making strategic investments, which have a financial return component, but also other priorities, It's a little bit hard to measure, um, results, and on the acquisition side, it's really, you can create as much value as destroy value with an acquisition, depending on how a target gets integrated.

AI assessment note: “the ability for a large company to disrupt oneself is quite difficult.”

Q You said about access there to the data itself. Do you think, then, for startups, there is a really significant data incumbency advantage to the likes of, say, your previous employer, Microsoft, or Google, or Facebook? Do you think this is a large and worrying data incumbency Um,?

A I do think that there are big data incumbencies, and I do think that they require, if you will, they're a challenge to overcome I do think that the barriers to overcome them are not so high. Um, as previously mentioned, the incumbents oftentimes get in their own way. So, um, but also I think the agility of a startup and, um, with execution and access to other sources of data. Right now, as we speak, companies that are actually being created and founded to enable this AI movement, to enable the, to provide data sets, To enable the training of algorithms for whatever purposes, whether it's, you know, to solve security problems or, um, marketing tech problems, sales tech, whatever the end application may be. And by the way, um, just sort of on the, on this whole notion of data sets, this is not a new problem. I mean, think about what has happened with ad, ad exchanges and ad networks and the aggregation of data. We've seen data marketplaces in the past.

AI assessment note: “I do think that the barriers to overcome them are not so high.”

Q I do have to ask, you mentioned one really interesting element for me there, and it was data sets, uh, and kind of the incredible amounts of data that's produced from all the devices. Do you think in terms of ownership, these will be in the form of Public data sets in the future, or will they be privately owned by the companies that produce the devices?

A So that's, that's a very fair, fair question. I think it will be a hybrid. I think the, there is a bubbling, if you will, consumer sort of reaction to consumer data being used by large entities and tech companies and the likes to these entities advantage and not for the consumer advantage. And you see all sorts of movements on the ad tech Do not track and whatnot. I think reality will be that data will remain proprietary to a certain extent. You know, you have entire ecosystems around Google and Uber and Facebook where they're effectively AI companies leveraging the large data sets, but you'll also see this push from consumers to have more control. So probably a hybrid where, and even the recent regulations that emerged and just there was a vote Um, I believe earlier this week around consumers having more power over their own data. So I think we might see a world where consumers have to buy in and have to approve and get some sort of benefit, whatever the case may be. That's somewhat contradicting with what we have seen with the millennials and the younger generations around this notion of privacy and data, them being a little bit less concerned. So I think you have a push and pull, but overall we will see a little bit more of a move, I think, in favor of consumers.

AI assessment note: “I think it will be a hybrid. I think the, there is a bubbling”

Q The next logical question for me then is for you as an investor, do you think these data sets that are privately owned are really a kind of motive defensibility for these companies, or do you think actually, you know, it's not a significant form of IP that would prevent you from investing?

A Listen, no single factor alone makes for a successful investment. Um, I think Forget data. Forget technology. In my view, you always start with people. They can make or break you. There is no artificial intelligence that replaces human execution. So, at the end of the day, sort of, you always start with, who are my founders? Who's my management team? And, you know, how can they translate vision to execution? Having said that, the more barriers to entry one has, the, um, the higher the chances of success. So, Having the right AI technologies and apply, you know, technologies and products is important. Having the right data sets to train these models and to create even more, um, not just training, but also to increase the data sets is important. How you get those data sets, you know, whether they're closed or open, it's also important. On the one hand, you might want to have proprietary data, but on the other hand, by doing right with consumers, you're creating a certain brand and a certain trust Between consumers and your product and your company, and that can be a barrier to entry and a brand differentiator in and of its own.

AI assessment note: “Forget data. Forget technology. In my view, you always start with people.”