Pichai says Google has found that two programmers working together do better than two working separately, a practice called pair programming.
“do something called pair programming. We have found that two programmers working together are better than them working separately.”
HBR IdeaCast · 0:02:01
Pichai says Deutsche Bank is a Google client that uses generative AI to give insights to its financial analysts.
“we have clients like Deutsche Bank is now using generative AI to give insights to their financial analysts.”
HBR IdeaCast · 0:02:36
Pichai says PaLM 2 was the model Google had in production and launched across its products at the time of the interview.
“The model in production, which we are using across our products and which we have launched is Palm two.”
HBR IdeaCast · 0:03:43
Pichai says Google's next state-of-the-art model, called Gemini, was being built by its new combined unit, Google DeepMind.
“the next state-of-the-art model we are working on, you know, with our new combined unit, Google Deep Mind is called Gemini.”
HBR IdeaCast · 0:03:43
Pichai says the next generation of AI models will be multimodal, trained on text, images, audio and possibly video.
“the next generation of models will be multimodal. Uh they're both trained on different modalities text, images, audio, maybe video”
HBR IdeaCast · 0:03:43
Pichai says the interview came in Google's seventh year as an AI-first company.
“this is our seventh year, you know, as an AI first company.”
HBR IdeaCast · 0:05:42
Pichai says Google had internally built a conversational system called LaMDA, based on large language models, before ChatGPT became public.
“internally, we had built what was called Lambda. Internally, we had built a conversational dialogue based on, uh, these large language models”
HBR IdeaCast · 0:06:35
Pichai says Google has been more conservative than some rivals in releasing generative AI because of the scale at which it serves users.
“as Google, we've been, uh, more conservative in our approach, given the scale at which we serve users.”
HBR IdeaCast · 0:08:04
Pichai says Google has built generative AI into more than 25 of its products, including Gmail, Google Docs, Search and YouTube.
“we have incorporated generative AI in over twenty five of our products, be it Gmail or Google Docs or search or YouTube and so on.”
HBR IdeaCast · 0:09:44
Pichai says Google uses red teams whose explicit goal is to break its AI models so weaknesses can be fixed.
“we have red teams whose explicit goal is to break these models in various possible ways.”
HBR IdeaCast · 0:10:35
Pichai says Google was still working on watermarking and metadata so AI-generated images can be identified and traced.
“still in the process of doing this work, but we are doing work to add watermarking and metadata.”
HBR IdeaCast · 0:11:05
Pichai says he previously called AI the most profound technology humanity will ever work on, more profound than fire or electricity.
“many years ago, I I called AI the most profound technology humanity, uh, is working on and will ever work on, uh, more profound than fire or electricity.”
HBR IdeaCast · 0:12:27
Pichai says the most important regulation the US could pass to help AI is a stronger privacy foundation, such as a national privacy bill.
“at least speaking from a US standpoint, you know, I think the most important regulation which we can pass, which will also help AI, is is a stronger privacy foundation.”
HBR IdeaCast · 0:15:04
Pichai warns that overly burdensome AI regulation would let only big companies comply and would stifle startups and the open source community.
“you can't make the uh regulations onerous. That means the big companies can do it, but, you know, you stifle innovation from startups or from the open source community.”
HBR IdeaCast · 0:16:07
Pichai says organizations can fine-tune base AI models on their own data so the models work better in their context.
“we call this fine-tuning. You can take these base models and in the context of your organization, you know, fine-tune it based on the data of the organization”
HBR IdeaCast · 0:17:20
Pichai says Wendy's was using generative AI so customers can order by voice at the drive-through.
“Wendy's has used generative AI so that people can, you know, use voice and as part of that drive-through, you know, order and and the system works that way.”
HBR IdeaCast · 0:18:23
Pichai advises companies to start generative AI adoption with four to five pilot ideas, driven from the top down.
“literally, this is about seeding your organization with four to five pilot ideas, challenging your organization from the top down”
HBR IdeaCast · 0:20:04
Pichai lists the pandemic, the war in Ukraine and rising interest rates as macro shocks the economy has absorbed.
“from the pandemic to the war in Ukraine and and to rising interest rates and so on”
HBR IdeaCast · 0:21:02
Pichai says Google's AI ambition covers four areas: improving knowledge and learning, boosting creativity and productivity, enabling others to use AI, and doing it safely.
“first is to improve knowledge and learning, second is boosting creativity and productivity, third, which has been important to us.”
HBR IdeaCast · 0:26:23
Pichai says Google's mission is to organize the world's information and make it universally accessible and useful, and that AI lets it pursue that mission more ambitiously.
“having a mission to organize the world's information and making it universally accessible and useful.”
HBR IdeaCast · 0:25:52
Pichai cites Google's work on quantum computing as one of its most ambitious moonshot projects.
“we are trying to solve quantum computing, uh, which is as moon shoty as it gets”
HBR IdeaCast · 0:27:39