Moonshots With Peter Diamandis: AI’s Real Risks, Big Wins, and Biology’s Leap

Peter Diamandis (0:01)

Good morning, and welcome to Moonshots Live, Peter Diamandis' inaugural event. Right from the start, the message is clear: this isn't a gathering built around fear, but around purpose, possibility, and action. Exactly. Peter tells the room that the mission is to stay hopeful and optimistic, to find a moonshot, and to help shape the future rather than wait for it. And he's got this great line: "AI is not a hand grenade; it is a jetpack." I love that framing. He argues that entrepreneurs are people who find problems and solve them, and he's pushing everyone to say their purpose out loud, because that statement could lead to a co-founder, a funder, or a partner.

But what about all the anxiety around AI? That seems to be the elephant in the room.

Right, so the conversation turns there pretty quickly. Peter and the panel reject the idea that super intelligence itself is the main threat. Instead, they argue the real danger is people using AI for malicious purposes, or systems being built with weak security and bad infrastructure. And Peter warns that a moral panic could set the field back 50 years. He says, "We are moving forward at warp speed."

There's also strong pushback against doomerism from major voices in the field. They criticize predictions like a 10 percent chance of societal destruction, calling them unscientific and irresponsible.

The argument isn't that AI has no risks, but that solutions should focus on trust, alignment, liability, and defensive co-scaling, not paralysis. Which brings us to the political fight over regulation.

Oh, this got heated. The group takes aim at Bernie Sanders' proposal for a permanent ban on artificial super intelligence, an immediate pause on advanced AI, and a new cabinet-level department. Peter calls it fear-mongering. He goes further, saying that capping intelligence is "the equivalent of communism in the 21st century." His biggest concern? That data center investment will flow to other states if policymakers overreact. But let's talk about the progress side. What's actually shipping?

The speed is remarkable. In just 22 days, there were 12 major model releases. Meta's Muse can book flights, order food, manage calendars, and shop. Claude Opus 5.5 is both cheaper and more capable. The larger point is that consumer AI may become the new interface for the internet. That's fascinating. And it extends beyond software. The discussion moves to robotics and embodied AI. A general-purpose model reportedly completed a driving cone course after one failed try, which the panel treats as evidence that specialized robot models may not be necessary. The idea is simple: drop a frontier model into a robotic body, and it can begin learning the physical world. But where this really gets exciting is biology.

Absolutely. Anthropic's lab work, CRISPR, digital twins, and the dream of solving disease all come into focus. The panel imagines cell-level models that can search for cures the way AlphaGo searched for winning moves. One speaker sums it up bluntly: "Biology is cooked." The episode ends with a sense that science, like software, is entering an era of rapid, measurable, and deeply consequential progress. The future is being created right now.