Siri, Humanistic AI
Join Mark for the latest episode of the Chief Executive Summit featuring Tom Gruber, co-creator of Siri and a pioneer in humanistic AI. This conversation dives deep into Tom's remarkable journey and the future of artificial intelligence, offering unique insights for leaders navigating this transformative era.Key Highlights:The Origins of Siri: Tom reflects on building Siri with a small team of 24 and the pivotal decision to join Apple, sharing personal anecdotes about working with Steve Jobs and
“Siri was the first mainstream AI experience for most people.”
“The VCs immediately told us, 'Don't sell, don't sell, don't sell. This is our 100X play, man.'”
“This is an opportunity for the founders to have impact on hundreds of millions of people. You don't get that chance more than once in life.”
“You basically have a choice when you're doing an AI project: you can either celebrate the machine learning as its own thing, or you can make a thing that says, 'Hey, this enables humans to do what humans love to do much, much better.'”
“A billion people in the world have untreated mental healthcare, and there's not enough humans in the world to sit down in a psychiatrist bench with a billion people.”
“AIs have no life experience, but they've read a million lifetimes of words.”
“In just five years, the social media companies addicted a billion to two billion people to use this stuff for one to four hours a day. It's the most effective addiction machine ever made in history, more than any casino or anything.”
Tom argues that once you name human impact as your real goal, rather than money or prestige, the downstream decisions about who to hire, what to build, and whether to sell change completely. The Siri acquisition story is the clearest proof: impact on hundreds of millions beat a 100X return.
Every AI project presents a fork: optimize the machine, or amplify the human. Tom insists this is an active leadership decision, not a technical one, and that most unintended harms, from social media addiction to biased outputs, trace back to leaders who never consciously made it.
The skills needed to delegate to AI, calibrate its strengths, monitor its outputs, and maintain critical judgment are the same skills good leaders use with human teams. The challenge is teaching those skills across the whole organization, not just at the top.
Tom draws a direct parallel between the KPIs executives set for people and the objective functions that drive AI models. Whatever you tell the system to optimize, it will optimize, so the accountability for unintended consequences sits squarely with the leader who set the goal.
Mental healthcare for a billion underserved people, and education that gives every child a personal tutor: Tom argues these billion-person problems could not be solved technically until now, and that the leaders who frame AI around those stakes will do the most important work of this era.
Answers come from this show’s transcript and nothing else. If Tom Gruber didn’t discuss it here, you’ll be told so rather than guessed at.