Alphabet CEO Sundar Pichai took the stage Wednesday at a Stanford event hosted by the university’s business school, offering some small insights into how he thinks about running one of the world’s most valuable tech companies.
It was a remarkable performance because Pichai has been on a rough streak lately. Google is widely seen as a latecomer to genetic AI, lagging behind Microsoft-funded OpenAI. That’s despite the fact that Pichai’s company has focused on artificial intelligence for most of the past decade, and Google researchers wrote the modeling paper on transformer models which really started the generative AI revolution. More recently, Alphabet’s Gemini LLM was criticized for creating wildly inaccurate depictions of historical situations, such as depicting America’s founding fathers as Black or Native American, rather than white Englishmen, suggesting an overcorrection for certain types of bias.
Interviewer, Stanford Graduate School of Business Dean Jonathan Levine, wasn’t exactly a hostile interrogator — at the end, he revealed that the two men’s sons had once played in a high school band together — and Pichai is adept at answering tough questions by posing them as further questions about how he thinks, rather than with immediate answers. But there were a few nuggets of interest during the talk.
At one point, Levin asked what Pichai tried to do to keep a 200,000-person company innovative against all the startups fighting to disrupt its business. It’s obviously something Pichai is worried about.
“Honestly, it’s a question that has always kept me up at night over the years,” he began. “One of the inherent characteristics of technology is that you can always develop something amazing with a small team from abroad. And history has shown this. Scale doesn’t always give you – the regulators might not agree, but at least running the company, I’ve always felt that you’re always sensitive to someone in a garage with a better idea. So I’m thinking, I’m thinking how do you move fast as a company? How do you have a risk-taking culture? How do you motivate for this? These are all things that you really have to work hard at. I think at least the bigger organizations tend to go bankrupt. One of the most counterintuitive things I’ve seen is that the more successful things are, the more risk averse people become. It’s so counter-intuitive. You often find smaller companies almost making decisions that bet the company, but the bigger you are, it’s true for a large university, it’s true a large company, you have a lot more to lose or you perceive that you have a lot more to lose. And so you find that you don’t take as many ambitious risk-taking initiatives. So you have to do it consciously. You have to push teams to do that.”
He didn’t offer any specific tactics that have proven successful at Google, but instead noted how difficult it is to create the right incentives.
“An example of that is how you reward effort and risk-taking and good execution and not always results. It’s easy to think that you should reward results. But then people start playing it, right? People take conservative things that you’ll get a good result from.”
It hearkened back to an earlier era when Google was more willing to take odd risks, notably showing the company’s ill-fated Google Glass. it didn’t work, but it was one of the first devices to experiment with augmented reality.
“WWe said recently, we went back to an idea we had at the beginning of Google for Google Labs. And so we’re setting up a thing where it’s easier to put something out without always having to worry about, you know, the full branding and burden of building a Google product. How do you get something out the easy way, the lightest way? How do you enable people to more easily innovate internally and get that out to people?”
Later, Levin asked what developments Pichai was most excited about this year.
First, he mentioned the multimodality of Google’s latest LLM — that is, its ability to process different kinds of input, such as video and text, simultaneously.
“All our AI models are already using Gemini 1.5 Pro. This is a 1 million context window and is multimodal. The ability to process massive amounts of information in any type of way on the input side and provide it on the output side, I think is mind blowing in a way that we haven’t fully worked out.”
Second, it highlighted the ability to link different discrete responses together to provide smarter workflows. “Where today you use LLMs as just an information seeker, but By combining them together in a way that you can deal with workflows, this will be extremely powerful. It could maybe make your billing system at Stanford hospital a little easier,” he joked.
You can see the full interview, along with an interview with Fed Chairman Jerome Powell that preceded it, on YouTube. Levin and Pichai start around 1 hour and 18 minutes.
