AI said so, but do you believe it?

AI adoption is happening at warp speed, but trust lags. A Stanford University survey found that 64% of Americans say that products and services using AI make them nervous. Pew Research Center says about half of Americans believe AI will worsen people’s ability to think creatively and form meaningful relationships. Even developers have doubts: More actively distrust AI tools than trust them.

In early October, Formidable brought together a group of AI innovators, investors, critics and thought leaders to discuss the implications of this fast-moving new technology. During a series of wide-ranging panel discussions, we delved into trust, bias, climate change, health care and many more topics. In this four-part series, we'll feature takeaways from these discussions. To view the full panels, please visit our YouTube page. 

Today we’re covering the panel “Trust, Verify, Save the World.” The panelists:

• Hansi Singh, PhD, CEO and co-founder of Planette AI, a San Francisco-based startup with an office in Seattle that launched in September 2022. It uses AI to predict long- and short-term term weather outcomes and has worked with NASA and the National Science Foundation.

• Michelle Pruitt, an AI innovation leader who was senior director of Business and AI Transformation at Microsoft until January 2025. She is now in the process of forming an AI startup.

• Steve Tapia, a distinguished practitioner in residence at Seattle University School of Law who has practiced entertainment, media and intellectual property law for more than 30 years.

Here are six key takeaways from the event.

Trust remains a major issue: Pruitt says she’s working with a pre-seed startup to develop “AI trust stops” focused on transparency, responsibility and user protection. “You have to build that discipline in organically from ideation, not as an afterthought.”

Treat AI as a business tool, not a friend: AI is not a person. It has no personality. There’s no need for niceties. “Don’t say please and thank you” (to AI), Pruitt says. “It’s not going to hate you. It’s going to work.”

AI has significant environmental costs but keep that in perspective: AI requires lots of energy, water for cooling data centers, and creates electronic waste from hardware. But Singh notes that AI models can also analyze data at the fraction of a cost of what it would take otherwise, especially when it comes to climate science and weather forecasting. “I think we just have this one umbrella called AI, and it means a ridiculous number of things. And, you know, usually people do not bother to disambiguate. It’s important to talk about those details and make those assessments of good use of AI or bad use of AI.”

AI is not a substitute for creativity: AI is only as good as the humans who create it. “We’re still going to need Beethovens and Mozarts. We’re still going to need Picassos,” Tapia notes. “I don’t know how many of you have played with the creation of art or music through AI. It’s kind of terrible. There’s always going to be that need for the creative spark that’s out there.”

Yes, AI contains human bias: A primary source of bias is data collection. The University of Washington last year released a study that said white-male associated names were favored by language learning models 85% of the time. Female-associated names were favored only 11% of the time, while white-male associated names were always favored over Black-male associated names. “Data really becomes a hiccup when you think about AI systems,” Pruitt says. “The biggest thing is to know is that these systems aren’t built to be biased, but our data is, and we have not yet figured out how to truly correct that. As humans, we have to apply our empathy and discernment.”

The future of AI has yet to be written: Remember the early days of the internet? Many thought it was a passing fad. Others thought it would be capable of doing almost anything. Both were false. “We can’t try to envision what’s going to happen,” Tapia says. “There are so many big issues nobody has a clue about.” Says Singh: “AI is not a single umbrella. There are many different types of use cases, different types of models. Let’s not put it all in a single bag.”