Sam Altman: ‘AGI in 2026’, just as Models [Mis]Train Themselves

A veteran engineering leader argues that high-performing technology organizations depend on customer closeness, small empowered teams and a culture that treats speed as a competitive advantage. Drawing on Akamai, Facebook and Microsoft, he describes how leaders can fight the organizational drag of growth while making room for consequential bets in the AI era.

As OpenAI heads toward an IPO, co-founder Greg Brockman is consolidating control over product, infrastructure and operations while Sam Altman turns toward strategy and fundraising. The reshuffle raises questions about executive turnover, enterprise ambitions, consumer hardware and the political costs of AI’s expansion.

Casey Moratori challenges the industry’s acceptance of sluggish software, arguing that performance begins with architectural choices, hardware literacy and theoretical limits rather than late-stage profiling. He also considers AI coding tools’ unsettled effects on engineering craft and autonomy.

Walleye CEO Will England makes the case for an AI-native hedge fund, from mandatory staff training to internal tools that process market data and meeting transcripts. He frames adoption as both a competitive necessity and a responsibility to workers navigating a changing economy.

Venture capitalist Bill Gurley makes the case that careers are built less by mythical mentors than by generous, candid peers. He and Molly Graham trace how reciprocal relationships, vulnerability and collective ambition can turn professional networks into engines of opportunity.

Anthropic’s universal culture interview turns every candidate, from engineers to marketers, into a test case for the company’s safety-first mission. The conversation examines how its rapid-fire questions, shared interviewer pool and insistence on intellectual honesty reinforce a distinctly cultish corporate identity.

Ryan Carson and Claire Vo discuss the new work of solo founders managing fleets of coding agents, from cloud-based development and automated PR reviews to customer triage and investor updates. For all the software AI can produce, they argue, product-market fit still depends on leaving the screen to talk with actual people.

GoFundMe CEO Tim Cadogan defends a for-profit model built on voluntary tips while confronting the platform’s role in medical crowdfunding, disaster relief and America’s frayed safety net. He also discusses AI fundraising tools, payment-processor constraints and the values behind decisions about which campaigns the platform will host.

Enterprise sales veteran Jen Abel maps the 15-step path from first outreach to signed contract, arguing that information advantage and careful project management beat rigid scripts. The conversation treats demos, pilots and procurement as chances to build internal champions rather than merely advance a pipeline.

Dan Shipper’s Every turns AI journalism, software experiments and a $20 subscription into a single feedback loop, testing what happens when a media company also builds the future it covers. He argues that automation does not eliminate expertise so much as push human work toward taste, judgment and new layers of management.

OpenAI’s claimed breakthroughs on long-standing mathematical problems have rattled a field suddenly confronting AI as both a powerful collaborator and a possible replacement. Mathematicians weigh the value of solved proofs against the human insight, training and new questions that make a discipline move forward.

Former Chrome engineering director Addy Osmani traces his path from building a browser as a rural Irish teenager to shaping DevTools and Core Web Vitals at Google. He also weighs how AI agents are remaking software work, leaving engineers responsible for judgment, verification and accountability.

Portola’s founders trace how an embodied AI companion became a storytelling experiment, with users and models co-writing relationships, memories and alien-world lore. The conversation considers AI less as a replacement for writers than as a new medium shaped by improvisation, responsiveness and human taste.

Casey Newton explains how an LLM-powered personal wiki turned years of reporting and daily research into a self-organizing archive, while demanding its own steady maintenance. The show also tracks OpenAI’s renewed safety measures and Meta’s mounting legal reckoning over youth social media harms.

Barry O’Reilly argues that artificial intelligence should sharpen leaders’ judgment rather than replace it, freeing them from administrative churn for deeper thinking and better decisions. The entrepreneur and author also warns that AI hype and content slop can shift unfinished work upward, demanding new standards of rigor, context and respect for colleagues’ time.

Molly Graham makes a case for career reinvention through calculated risk, beginner’s humility and a willingness to endure the emotional chaos of change. The new WorkLife host considers how small acts of courage can build resilience in a precarious job market and at midlife.

Product recruiter Chris Lee explains why job seekers should tailor their LinkedIn narratives to a specific target, treating the search as a test of candidate-market fit. He also walks through how recruiters scan profiles for stage, domain, seniority and hands-on AI experience before deciding whom to contact.

Semafor editor-in-chief Ben Smith makes the case for an anti-scale news business built on elite audiences, global events and aggressively independent reporting. He and Nilay Patel spar over creator economics, advisory boards, AI tools and whether audiences can still recognize journalism that refuses to be bought.

OpenAI’s Ian Silber argues that designers rattled by AI’s rapid advance are confronting an uncertain role, not an obsolete one. He describes a field where curiosity, systems thinking and a willingness to ship imperfect ideas matter as much as pixel-level polish.

Open-source developer Simon Willison traces how coding agents have turned ambitious software experiments into morning chores, while making judgment, review and cognitive capacity newly scarce resources. He and Claire Giordano consider the pressure this velocity puts on maintainers, contributors and the next generation of programmers.