The Moonshots mates run their first live AMA, taking audience questions on Zoom instead of the usual format. Topics range from AI consciousness and what comes after transformers to very practical career advice: mid-career professionals leveraging AI, young founders raising hardware capital, selling proprietary data to frontier labs, and rethinking education and parenting for a post-AGI world. Peter, Alex, Dave and Salim also field questions on BCI for disabilities, postfusion energy, planned longevity communities, and whether there's room for God in an AI-driven future. The tone stays optimistic but acknowledges a rough multi-year transition ahead as institutions get rebuilt.
Alex Wissner-Gross lays out two theories of consciousness (emergent complexity vs. brains as a 'localized antenna' tuning into consciousness) and says AI could achieve it under either model; Salim calls it a century-scale open problem.
Next-gen interfaces: personal AI filters and BCIAI▶ 7:13
Salim describes two coming interface shifts: a personal AI that filters/summarizes better than you can yourself, and brain-computer interfaces that could deliver preformed concepts directly to the brain, effectively adding a 'third corpus callosum.'
Fountain Life longevity-clinic franchisingLongevity▶ 12:32
Peter says Fountain Life (5 centers built, 2 more coming, plus a resort partnership) plans to franchise its operating system and AI diagnostics within roughly six to nine months, including an eventual center in Greece.
AI-era playbook for mid-career professionalsEconomy▶ 14:56
Dave Blundin advises experienced professionals with families and mortgages to watch for AI-lab-provided playbooks (franchise and data-collection deals), build small AI side projects for cash flow, and deliberately network with the five people they want to become.
Selling proprietary domain data to frontier AI labsAI▶ 20:38
A civil engineer with 30 years of project records asks how to get AI labs to actually close a data deal; Dave explains labs would rather write a large check than have a meeting, so the key is packaging the data for an easy yes and, if stuck, showing up in person.
AI regulation strategy in the public sectorGeopolitics▶ 25:19
Alex argues the real risk right now is overregulation, not underregulation, given the stakes of a superintelligence race with China and 150,000 daily deaths that better medicine could address; Dave adds that declaring yourself the in-house expert on a coming risk gets you pulled into the response team when it hits.
Fundraising advice for young hardware foundersEconomy▶ 32:08
Dave tells two 18-year-old founders that hardware VCs now want to see a gelled founding team, enough capital to reach a working prototype, and evidence of prior builds; Alex adds that framing the company as an 'intelligence company whose products happen to have atoms in it' matters more than the hardware label.
Engineering serendipity: building people-connecting communitiesOther▶ 35:59
Salim and Peter describe efforts (Moonshots Live events, a possible community bulletin board, an early project trying to measure and algorithmically engineer 'luck') to connect like-minded builders so they can find their own Steve Jobs / Woz-style collaborators.
Rethinking K-12 education for a post-AGI worldAI▶ 45:03
Responding to a teacher who says classrooms still look like the 1960s, Salim argues fixed curricula are obsolete since a child with AI can learn a day's classroom content in an hour, and pushes 'demand-side' education: find a problem you care about first, then learn the skills to solve it, alongside the '6 Cs.' Peter points to TKS and Alpha School as early models.
BCI and AI for people with disabilitiesHealth▶ 49:31
Answering a retired social worker, Alex recalls an MIT quadriplegic classmate who typed with a pencil in his mouth and says AI-driven communication tools would now let him be fully functional; both Alex and Peter say BCI breakthroughs will help people with motor, neurological and psychiatric conditions broadly, not just narrowly-defined disabilities.
AI-driven energy efficiency and the race beyond fusionEnergy▶ 53:41
Alex says AI is simultaneously improving compute-per-watt efficiency and being used to hunt for entirely new, breakthrough energy sources; he expects a 'postfusion' era of energy within about ten years, discovered by AI rather than sourced from space.
What replaces LLMs and transformers nextAI▶ 1:12:38
Alex distinguishes LLMs (the next-token prediction task, which he thinks has near-infinite runway) from transformers (the current architecture, already being replaced piece by piece); Dave predicts compute moves to photonics within 18 months to two years, after which AI reinvents physics and 'explains everything else to us.'
Navigating the 'period of pain' before abundanceEconomy▶ 1:17:13
Peter says we're already mid-transition, with roughly 50 core institutions unable to survive AI plus ~20 other simultaneous 'Gutenberg moments'; Salim expects the turbulence, underway for 5-10 years already, to run another 10-20 years, and both frame the goal as minimizing the trough rather than avoiding it.
Religion, simulation, and the case for an ultimate intelligenceOther▶ 1:23:28
Asked whether they're religious or believe in an intelligently designed universe, Salim says either we're in a simulation (making the simulator the 'all-seeing intelligence') or humans simply invent gods to cope with uncertainty, and that as evidence accumulates the two live options collapse to 'no god, just a tech-driven future' or 'it's a simulation and doesn't matter either way.'
Predictions made
openDave Blundin: This year will bring on the order of 10 to 100 opportunities where AI labs directly hand people a franchise or data-collection playbook to get involved and succeed.
“I suspect that this year there'll be on the order of maybe 10 to 100 opportunities where the playbook is handed to you.”
Your call:
openPeter Diamandis: Fountain Life will open its franchising model (operating system and AI diagnostics) to outside interest within roughly six to nine months.
“I think that's probably at least six to nine months away from being an open call for interest.”
Your call:
openAlex Wissner-Gross: Within about the next 10 years, we will be well into a 'postfusion' era with new energy sources discovered by AI, beyond fusion, fossil fuels, solar and wind.
“I would fully expect in the next 10 years we're going to be well into our postfusion era.”
Your call:
openDave Blundin: The compute substrate will move to photonics, with the whole current AI infrastructure running millions of times faster within roughly the next 18 months to two years.
“I am 100% sure that the compute substrate is going to move to photonix... that's just in the next really 18 months to two years is my guess.”
Your call:
openPeter Diamandis: There will be a roughly two-to-eight-year period of elevated civil unrest risk, driven mainly by young men economically displaced or unable to start families, before the institutional transition settles.
“There's a period of you know two to eight years that I talk about concern of civil unrest.”
Your call:
openSalim Ismail: The current period of institutional uncertainty, already underway for 5-10 years, will continue for another 10 to 20 years before peaking and receding.
“It's going to go on for another 10 to 20 years, it'll go like a peak and then it'll [recede].”
Your call:
Numbers that matter
88% of Fountain Life patients screened have detectable coronary artery diseaseCited by Fountain Life's chief medical officer during the sponsor segment on CT angiography with AI analytics.
23% of those patients had soft plaque invisible to standard calcium-score CT scansUsed to argue for AI-driven multimodal cardiac screening over calcium scores alone.
50% of people who die of heart attacks have no prior warning signsFramed as the case for prevention-first, AI-assisted early detection at Fountain Life.
More than 150,000 people die on Earth every dayAlex Wissner-Gross's argument that overregulating superintelligence delays medical breakthroughs and costs lives.
$250 billion has flowed into Louisiana's data-center buildoutAn audience member (a deep-sea-robotics engineer turned AI-native cooling specialist) describing SpaceX-driven investment in his state.
The Exo community has grown to about 50,000 peopleSalim Ismail describing a trained 'peace corps' for helping institutions transition during the turbulent AI era.
The human brain has roughly 100 trillion synaptic connections and 100 billion neuronsSalim Ismail using brain scale to frame the potential of BCI as a 'third corpus callosum.'
Worth digging into
🕳️ AIXI-style 'superbayesian' algorithms trained without human inductive priors
Alex Wissner-Gross name-checks (garbled by auto-captions as 'AXY') a class of algorithms for building intelligences with a near-universal prior, offered as the answer to whether AI is trapped by human self-limiting beliefs — a real, decades-old theoretical framework (Marcus Hutter's AIXI) worth grounding against what he's actually claiming.
🕳️ Mercor as a channel for monetizing proprietary domain expertise
Dave Blundin recommends mercor.com as a starting point for professionals with valuable domain knowledge to find AI-lab buyers, ahead of more specialized data marketplaces he expects to emerge.
🕳️ TKS (The Knowledge Society) as a purpose-first education model
Peter Diamandis points to TKS as a live alternative to curriculum-based schooling for ages 13-18, built around helping teens find purpose and become entrepreneurs — directly relevant to the education-overhaul discussion.
🕳️ 'Remember Me' — AI-agent life-story interviews for aging parents
Peter describes an internal project where AI agents interview an elderly parent to build a biography and near-functional avatar, calling the avatars 'almost good enough' — a concrete near-term application of the psychotherapist-guest's own idea.
🕳️ Prediction-market benchmarks testing AI foresight on future world events
Alex claims prediction markets and benchmarks already test whether AIs with a fixed knowledge cutoff can forecast subsequent real-world events, offered as evidence against the 'AI just parrots training data' critique.