Google CEO, Sundar Pichai, speaks out on AI and Google Bard

Photo by Mojahid Mottakin on Unsplash

Sundar has been speaking about AI for years. However, he has been muted during this recent AI hype cycle while Microsoft and OpenAI receive all the attention. On March 31st, 2023, Sundar Pichai spoke with the NY Times on their “Hard Fork” podcast and shares his current views on AI. The podcast was different than I expected. I expected Sundar to heavily promote Google’s Bard, which was recently released. Instead, Sundar talked about Bard’s limitations and explained his reasons for moving slowly. Also, contrary to popular belief, Sundar did not order a ‘code red’. While we may never know who ordered the ‘code red’ within Google, Sundar is now harnessing Google’s resources to move forward with urgency.

Did you order the code red?!

Listen to the Podcast: Apple | Spotify | Google

Google C.E.O. Sundar Pichai on Bard, A.I. 'Whiplash' and Competing With ChatGPT

Below is my summary of the podcast for those who need the tl;dr.

• Reception to Bard has been lukewarm since its release.
• Developing narrative:
  — Google somehow fell behind in the generative AI race despite having a dominant position for many years and creating the Transformer, the foundation for ChatGPT and other programs.
  — A lot of Google’s researchers left to form their own AI companies or went to competitors.
  — Google is hamstrung by big company politics and bureaucracy.
• Google is facing the classic “Innovator’s Dilemma problem”. How does Google capitalize on generative AI without destroying its search advertising business?
• The current version of Bard is limited since it uses a lightweight and efficient version of LaMDA — “I feel like we took a souped-up Civic, kind of put it in a race with more powerful cars.”
• Google does not want to use more capable models in Bard yet but it can plug in more capable models over time.
• Is there a real existential risk or a real generational opportunity for Google?
• Imagine if everyone had a personal assistant? How life changing would that be?
• Bard’s strength is asking it questions that spark your imagination and where you can fact check. Ie.
  — What should I do with my dad on his 80th birthday?
  — What movies can I watch on a Friday night?
• AI expands the possibilities of what you can do so you can do more. Models will become more capable and Google will follow the user journey. People will evolve over time and adjust their behavior to what the models can do.
• Sundar is harnessing Google’s resources to move forward with urgency and asking teams to understand how AI will translate into deep, meaningful experiences. Sundar wanted Google to think in an AI-first way since 2015.
• Sergey is hanging out with Google’s engineers and he says AI is the most exciting thing he has seen in his lifetime.
• Work around privacy, safety, responsible AI is more important than quickly getting products out the door. Google is taking its time to be very responsible.
• AI is the most profound technology humanity will ever work on.
• 1000 technology leaders and researchers, including Google employees and Elon Musk, signed a letter asking to pause training LLMs more powerful than GPT4. Sundar believes the spirit of this letter is worth being out there.
• Google has been very clear about responsible AI and is one of the first companies to put out AI principles and to issue progress reports.
• Responsible AI is analogous to when the power of recombinant DNA technology was introduced in the 1970’s; experts got together and started thinking about frameworks.
• Sundar has never seen a technology in its earliest days have as much concern about AI which gives him hope. Lots of important people are very concerned which gives Sundar hope that humanity will rise over time and tackle what needs to be done.
• “The way you get things wrong is by not worrying about it.” If you don’t worry about something, you will be surprised.
• Two categories of AI risk that people are worried about:
  — Short-term worries: chat bots get things wrong or provide biased or bad answers.
  — Long-term worries: AI destroys civilization.
• Sundar’s Whiplash — people want him to move faster, compete, and not let Google’s innovations go to waste but what if something goes really really bad?
• Programming will become more accessible to more people and will evolve to a more natural language way of programming over time.
• How does the web survive if Bard can answer questions without directing people to another website?
• Part of the reason Google is careful with Bard is because it wants to engage with the publisher ecosystem and work together to evolve the ecosystem.