The Future of AI: A Conversation with Sundar Pichai
Recently Sundar Pichai was interviewed by Jonathan Levin at Stanford Graduate School of Business.
Below is the transcript of that conversation.
Jonathan Levin: Sundar will be in conversation with Dean Levin, so please join me in welcoming Sundar Pichai to the stage.
Great. So Sundar, welcome, and I’ve been looking forward to this. Thank you for being here.
Sundar Pichai: Well, first of all, it’s good to be here. Thanks for having me.
Jonathan Levin: Okay, so I’m going to talk to Sundar. I’ve got a bunch of topics for discussion today. We’re going to talk a little bit about Sundar’s leadership of Google, a bit about the political issues facing Google and Alphabet, and of course I want to ask him about AI…
Sundar, to start off, though, you’ve been CEO now of Google for almost a decade, actually, and I wonder, as you look back at that time, particularly going back to the start, what are the things that have sort of gone as expected and the things that have gone differently?
Sundar Pichai: Well, probably the biggest thing is…the scale at which things are working…So I think the scale of change, when you’re in it, it’s, you know, it feels like you’re going through a lot, but if you fast forward a decade from now, you would have kind of normalised on the change, and you would be thinking.
Jonathan Levin: Let me ask a question about Google’s own growth…How do you take a, how do you, the company is, everyone thinks that it’s being hugely entrepreneurial and innovative, and it is. How do you do that at the scale you’re talking about? What are the things you have to do to keep a 200,000 person company being innovative?
Sundar Pichai: It’s a great question…For all, you know, one of the inherent characteristics of technology is, while people always think it’s not true, you can always develop something amazing with a small team from the outside, right? And history has shown that, right? And so scale doesn’t always give you…you’re always susceptible to, you know, someone in a garage with a better idea…
Jonathan Levin: Anything that works particularly well? Are there tactics you use or strategies that have been especially effective?
Sundar Pichai: You know, an example, maybe I would talk about it in the concept of AI now. You know, how do you create capacity in the system?…And so we are setting that thing up by which it’s easier to put out something without always worrying about, you know, the full brand and the weight of building a Google product.
Jonathan Levin: I’m gonna come to the AI piece in a minute, but one other question about scale. So as a $2 trillion company almost, and with it touching people’s lives, not just in this country, around the world, I wanna ask you about sort of the responsibility the company has. So Google, you famously had the motto, don’t be evil. What does that mean?
Sundar Pichai: Look, it simply meant, you know, do the right thing, right? And, you know, at a company of our scale, you’re constantly making decisions at all levels of the company, right? And when you make these decisions, you know, different people interpret it differently, but to me, it always meant, you know, always try to do things that are positively impactful…
Jonathan Levin: So for a long time, the company got to run largely outside of regulatory control. And of course, that’s definitely not true today…Is it gonna be possible going, you’re gonna have to, many things you’ve had to specialise, things around privacy or data and stuff like that. You have to do things differently in different countries. Is that, how do you run a business that has to respond to all these local environments and different rules, different political environments, and also have it be a global company that takes advantage of all of the scale?
Sundar Pichai: It’s definitely complicated…From the Google standpoint, you know, I viewed it as we’ve always, it’s been very clear to me that, you know, we’re going to comply with the laws in all the countries we operate in…And, you know, in fact, the way I think about AI is what part of AI will make it easier for us to, you know, to comply with regulation at scale.
Jonathan Levin: So speaking of AI, I mean, you’ve been at the, including DeepMind and Google, you’ve been at the forefront of a lot of AI for, actually for, well, many years now. Of course, things have gone very fast in the last 18 months. Did any of that surprise you?
Sundar Pichai: Oh, you know, for sure. Look, I, yes and no in a sense…And having said that, you know, maybe the pace at which society would adopt it, maybe I probably had a different sense of the trajectory in mind. And so all of us, I think, you know, positively surprised us at the scale at which people are engaging with the technology…
Jonathan Levin: So when Google was started, the mission that was articulated was to organize the world’s information. And back in the day, that meant presenting people with an array of links and ways to navigate the Internet, find information, find answers. And now, these AI engines, you just give them an answer. How are things going to evolve? Like, how will people access information in 10 years?
Sundar Pichai: Look, I think it’s one of the common myths around that Google has been 10 blue links for a long time…So to me, I view it as a more a natural continuation. Obviously, with LLMs and AI, I think you have a more powerful tool to do that, a powerful tool.
Jonathan Levin: So here’s a hard question though that I think comes from that…When you give an answer, you just give one answer. And there’s a lot of discretion in terms of setting the argument, how the answer is…How are you going to manage that? And what, how do you think about what answer to give to questions, particularly questions that have some, that might be value laden in certain ways?
Sundar Pichai: Look, I mean, let’s talk about search for a second here…And so, I think that’s different from when you’re in a chat bot, and I think that’s the more active area of research, where sometimes it has its voice. So how do you get those moments right? And, you know, again, for us, I think it’s an area where we will be deeply committed to getting it right.
Jonathan Levin: That’s fascinating. And would you, can you envision a world, how will you make those decisions? Like, those are decisions that, this is basically the way, in the future, I mean, if Google retains its market share, you really are the way that, you know, I don’t know if the majority of the world, but a pretty big share of the world’s population, you’re the first source for learning anything. Is the executive team going to make those decisions about whether at the end of the day, are you going to have some outside way? Do you think regulators are going to come in? How do you see that evolving?
Sundar Pichai: Right. Today, you know, there are many products. Take YouTube, for example. We have community guidelines for YouTube…But, you know, we do it. We regularly consult with a wide range of groups. And in the case of search, again, we have our search rater guidelines. It’s a public document. Our raters are based around the world, you know, around the world, in the U.S. and all part. We statistically, you know, do it in a way to represent the viewpoints of our user base, right?
Jonathan Levin: When you, the architecture that everyone is using to do these large language models, which is mind-blowing…The architecture came out of Google. There was a paper written by a bunch of people within Google in 2017. They’ve almost all left, actually, to go start other companies. But it was a competitor that came first to market with a consumer product. Does that make you think differently about how you keep your ideas inside the company versus just putting them out into the world?
Sundar Pichai: Look, I think the amazing thing about technology is all of us, at any given point in time, are building on top of innovations which are happening…I think that approach has served us well over many, many years. I am super confident in AI because of that kind of work…And so, I think very long-term, and I think that approach, the fact, I’m proud of the fact that we, Transformers, MixtureFX, so many of the underlying technologies were developed at Google and will continue to be developed at Google.
Jonathan Levin: Here’s one more, which is that if you go outside of the model or it’s from you. I’m just trying to… This might have been from an attendee, actually. So, we’re trying to do the array of different perspectives approach here.
So, another, one of the other technologies, actually, may be one of the most exciting. Developments came from DeepMind when they developed AlphaFold to predict protein structures. Are you gonna become a drug company now that you have developed this technology?
Sundar Pichai: Again, AlphaFold is an example of one of the deepest potentials for AI. Its ability to actuate science and discovery is going to be very profound. AlphaFold has done what would typically take a PhD student years to do, but it has done it for 200 million proteins in just three months. It’s one step in the process of drug discovery, but the number of biologists and researchers using AlphaFold has increased significantly.
We have given a database in AlphaFold and we will partner with pharmaceutical companies to bring AI and technology to help with drug discovery. We also have a company called Isomorphic that is further developing the AlphaFold technology. We are committed to finding every possible way to actuate progress in this field.
As for self-driving cars, the progress in technology has been remarkable. Driving in complex cities like San Francisco has shown significant progress. In the next six to 18 months, there will be even more progress. Having computers assist humans in driving will become the norm in the future.
It is clear that AI will have a positive impact on many areas, including healthcare. However, the potential labor market effects of AI are still uncertain. AI is an immature technology that can make errors, so it will be a productivity booster in many professions. It will create new classes of jobs, just like automation has done in the past. We have to keep an open mind and work together to harness the benefits of AI while addressing any societal disruptions.
Maintaining trust is crucial for Google. People come to search for information at vulnerable moments, and we have to earn and maintain their trust. We have to be responsible in deploying AI and listen to feedback from users.
The most exciting thing we will roll out in the next year is the ability to process huge amounts of information in any modality and tackle workflows. It will be powerful and applicable across society.
My thoughts on this conversation — specific to points highlighted by Sundar that correlates to Search Engine Optimisation because I’m SEO practitioner and those points definitely affect my career and day to day workflows.
