Across AI sources
What researchers, builders and critics are saying about AI.
Their arguments, examples and disagreements, with links to the original material.
A Moonshots listener whose career spans chip manufacturing and video games asked where tomorrow’s creative partnerships will begin. Salim Ismail described an early experiment in using AI to connect people through less obvious affinities. John Hagel’s writing on open-ended opportunities helps explain the idea—and why a surprising introduction is only the start.
At a Moonshots AMA, Griff D. proposed having AI models explore competing pictures of reality and judge them by their predictions. Alexander Wissner-Gross argued that scalable learning and search already offer a route beyond human assumptions. Their disagreement connects a practical experiment to Sutton’s Bitter Lesson, the theoretical agent AIXI and OpenAI’s claimed mathematical discovery.
A Moonshots caller asked for an alternative to the 22-year-old founder working 80-hour weeks. The panel’s suggestions offer a starting point: plan around your actual obligations, protect experimentation time, build something narrow and find receptive peers and customers. Expert contracting is another route—but neither income nor ready-made business opportunities are guaranteed.
An educator asked the Moonshots panel why classrooms of five- and six-year-olds still look like the 1960s. Salim Ismail proposed organizing learning around problems children care about, rather than future job titles. The discussion connected personalized pacing, purpose and a lifelong AI tutor—but left questions about practice, values and school’s social role.
After more than eight years caring for three loved ones, Tugboat Caregiving founder John Regan asked the Moonshots panel how to reach families already in crisis mode. Their suggested starting point was everyday planning: describe a day, ask an AI assistant what could change, and choose a useful next step.
A Moonshots listener named Scott described bringing a background in diving, underwater robots and electronics to a Louisiana business focused on data-center cooling. His account offers a concrete route into AI infrastructure: the equipment that carries heat away from servers, and the industry standards that help it fit together.
A civil engineer said Mercor and Scale had shown interest in his company’s 30-year project archive, but weeks had passed without a deal. Dave Blundin’s advice focused on packaging, pricing and personal access to buyers. Mercor’s published licensing process offers another route—without selling the company or the archive outright.
Asked how trusted local doctors could bring Fountain Life's longevity approach abroad, Peter Diamandis outlined a franchise model built around its clinical operating system and AI tools. He estimated that an open call for interested partners was probably at least six to nine months away from the remarks—not a timetable for opening centers.
Emad Mostaque proposes locally owned companies that would sell AI services, deploy robots and reserve a stake for young people. The aim is to give communities ownership of productive assets before automation weakens wage income. Here is how the proposed businesses and financing would work—and what remains unresolved about benefits, control and risk.
After briefing an oil company's executives in Houston, Salim Ismail offered a test for distinguishing AI adoption from organizational change: take the tools away and ask which workflows would change. His proposal shifts attention from better assistants to redesigned operations—and raises questions about useful data, reliable evaluation and dependence on suppliers.
Companies save us from negotiating every task on the open market. What happens if AI agents make those negotiations cheap? The proposed “Coasean singularity” points toward smaller firms built around networks of people and agents—but privileged AI access, platform power and control of supply chains could pull in the opposite direction.
OpenAI’s September 1 announcement lets healthcare organizations connect authorized Epic patient records and public healthcare sources to ChatGPT. On Moonshots, Peter Diamandis welcomed a conversational layer over a system he calls “a bear to navigate.” He and Emad Mostaque then argued for a bigger step: routine AI cross-checks of clinical decisions.
Follow a thread