Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Sundar Pichai said AI could be “very harmful” if deployed wrongly, and that its fast development did keep him up at night. He made the remarks in an interview with Scott Pelley for CBS’s 60 Minutes, published April 16, 2023—not as a new warning in 2026. His concerns included disinformation, confident errors, uncertainty about how models behave, and whether society could adapt quickly enough.
What did Sundar Pichai say keeps him up at night?
In the 60 Minutes interview, Pelley asked Pichai about the risks of AI. Pichai replied: “If deployed wrongly, it could be very harmful,” then added, “The technology is moving fast. Does that keep me up at night? Absolutely.” ITPro reported the exchange on April 17, 2023; the CBS transcript of the interview provides the broader context.
His remark was conditional: he was concerned about what could happen if powerful AI were deployed carelessly or misused. It was not a claim that every AI system is inherently harmful, nor a technical forecast with a stated timetable.
Why was he concerned about AI?
Pichai’s worry joined several related risks: systems can produce plausible but false answers, AI can make fabricated media more convincing, and developers do not always fully understand why a model behaves as it does. He also worried that technology might advance faster than social institutions can respond.
#1 Best Overall
AI may outpace society’s ability to adapt
Pichai described a mismatch between the pace of technological change and the time institutions need to think and adapt. “On one hand I feel, no, because you know, the pace at which we can think and adapt as societal institutions, compared to the pace at which the technology’s evolving, there seems to be a mismatch,” he told Pelley.
He also said he was encouraged that people were discussing the implications early. His answer combined concern about readiness with the possibility of addressing the risks before the technology developed further.
Rank #2
Models can produce confident errors
The segment showed Bard producing titles for books that did not exist. Asked about hallucinations—plausible-sounding but inaccurate outputs—Pichai said, “No one in the, in the field has yet solved the hallucination problems. All models do have this as an issue.” That was his assessment in April 2023, not a statement about the present-day performance of every AI model.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →AI could amplify disinformation
Pichai warned that AI could make disinformation much more scalable. He described the possibility of creating a convincing video of someone saying something they never said: “And it could look accurate.” The example illustrates why fabricated media can be difficult to assess by appearance alone.
Some model behavior can be hard to explain
Pichai called this the “black box” aspect of AI: “You know, you don’t fully understand.” The interview also discussed a Google system that reportedly translated Bengali after limited prompting, despite not being trained to know it. That was an example presented in the segment, not proof that all models behave this way or a settled explanation of how such behavior arises.
Jobs and tasks may change
The discussion distinguished jobs disappearing from the work within jobs changing. Google senior vice president James Manyika said, “Something like more than two-thirds will have their definitions change. Not go away, but change.” That figure was Manyika’s claim in the 2023 interview; the transcript does not provide a methodology or independent labor-market analysis for it.
Pichai separately named knowledge workers—including writers, accountants, architects, and software engineers—as people whose work could be affected. These comments describe anticipated change, not a definitive prediction of current job losses.
What did Pichai propose doing about the risks?
Pichai argued that AI development should involve people beyond engineering. In the interview, he said, “This is why I think the development of this [AI] needs to include not just engineers, but social scientists, ethicists, philosophers, and so on.” He also said, “It’s not for a company to decide.” His point was that choices about AI’s development and use call for broader participation, not decisions made by one company alone.
Best Value
At the time, he described Bard’s release as a limited experiment. Google intended to deploy it gradually, gather user feedback, and build more robust safety layers before releasing more capable systems. This describes the approach discussed in 2023; the interview does not establish Google’s current product or safety practices.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to read the interview’s claims
The segment combined demonstrations, interviewees’ interpretations, and predictions. A demonstration shows what happened in that particular instance; it does not by itself establish how common the behavior is. Likewise, the employment estimate belongs to Manyika, while the comments about risk and governance belong to Pichai. Keeping those distinctions clear prevents a single interview from being mistaken for an independent assessment of every claim.
Pichai’s shortest summary of AI’s reach was: “AI will impact everything.” In context, his concern was not only about what the technology could do, but whether it would be deployed responsibly and whether society could keep pace.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
