September 2026
Philip Zelikow of the Hoover Institution argues that U.S. leadership in AI will depend not only on chips and software but on the ability to build electricity generation and transmission at far greater speed. America’s fuel resources and nuclear prospects offer little strategic advantage, he says, if projects take 10 to 15 years while China expands power capacity across multiple sources at scale. Zelikow calls for bipartisan regulatory changes that turn energy abundance into operating capacity for a data-center-intensive economy.
NVIDIA argues that AI-factory capacity is constrained by the facility as a whole—grid connection, power delivery and cooling—not GPUs alone. Its DSX AI Factory Platform combines grid-responsive demand management, DSX MaxLPS power controls, 800 VDC distribution and 45°C liquid cooling to direct more of a fixed power budget to compute. NVIDIA says MaxLPS can enable up to 40% more compute within the same power budget.
Chris Williamson argues that men’s vulnerability is often accepted only when status, success or physical strength has already established them as masculine; men who appear isolated or unstable are more likely to be ridiculed than supported. Psychiatrist Dr Sasha Hamdani says boys taught to suppress sadness and sensitivity can come to rely on anger as one of the few permitted expressions of distress. Both distinguish genuine composure from emotional concealment, and argue that sensitivity can become either emotional intelligence or self-directed shame depending on how it is handled.
Neil Zeghidour, co-founder and CEO of Gradium, argues that most real-time voice agents remain “walkie-talkies”: however quickly they respond, they can only listen or speak, not handle the overlapping signals that make human conversation work. Full-duplex speech models can represent simultaneous talk and backchanneling, he says, but they still sacrifice reasoning and tool-use capability relative to text-based agents. His proposed answer is a split architecture in which a lightweight speech model manages the live conversation while a separate text model handles harder reasoning and actions.
ElevenLabs presents its MCP for Claude as a way to manage AI ad production through chat while retaining access to the underlying ElevenCreative Flows workflow. In a tutorial demonstrated by Alec Wilcock, users can start with a product link, approve stills and voice takes, revise motion or narration, and reformat a finished ad, while Flows exposes the separate image, video, voice and composition steps. The process still depends on user decisions—and on manual product details when a linked page cannot be retrieved.
MIT Sloan professor Paul Osterman argues that roughly 35% of U.S. workers occupy “disposable” jobs—through contracting, organizational freelancing, or marginal W-2 employment—that leave them outside a durable employer commitment to training, benefits, and advancement. In his account, firms use these arrangements to cut costs and preserve flexibility, while contractors and marginal employees report lower earnings, satisfaction, and attachment to their organizations than standard employees. Osterman argues that training can improve individual options, but changing the system requires stronger classification rules, labor standards, benefits, and bargaining power.
OpenAI developer-experience lead Charlie Guo argues that voice agents should not be designed to answer every spoken request with speech. The underused model, he says, is speech to action: a user states an intent, the system calls tools and changes the interface or completes a workflow, using audio only where conversation, hands-free operation or escalation warrants it. For developers, Guo’s practical case is that many of those tools already exist as application endpoints, handlers and workflows.
Anthropic is preparing to watermark future Claude output by subtly biasing its word choices, creating a statistical pattern that authorized detectors can identify even though readers cannot see it, according to Two Minute Papers host Károly Zsolnai-Fehér. He says the signal survives copying and limited editing but can be removed through wholesale rewriting; Anthropic’s own materials say detection indicates likely Claude involvement, not a particular user, chat, or degree of authorship.
Nvidia chief executive Jensen Huang argues that AI policy should target demonstrable failures at frontier labs rather than catastrophic forecasts he calls ungrounded. In a discussion joined briefly by President Donald Trump, Huang says US leadership will depend less on owning every important model than on deploying AI broadly through open and closed systems, compute, power, data centers and industrial capacity. He also contends that “superintelligence” already exists in bounded applications such as autonomous driving and protein science, making practical deployment—not speculative thresholds—the central challenge.
Dario Amodei’s call to “pace the frontier” would place outside evaluators and government-backed co-operation inside a competitive AI race, raising questions about antitrust, incumbent advantage and whether restraint can be made credible internationally. John Coogan says the plan seeks institutional safeguards across frontier labs and coordination with China, while David Sacks argues that companies concerned by their own models can slow development voluntarily rather than seek Washington’s approval for collective action. Gavin Baker, relayed by Jordi Hays, presents documented duty of care and third-party review as a possible source of accountability in future liability cases.
David Sacks, chair of the President’s Council of Advisors on Science & Technology, argues that Anthropic and OpenAI should slow or redirect frontier-model development if they judge their systems unsafe, but do not need new regulation, antitrust exemptions or liability waivers to do so. Speaking with Bloomberg’s Ed Ludlow, Sacks says existing legal exposure, customer demands and ordinary product responsibility should compel safer development, while transparency and independent audits can provide oversight. He warns that a mandated U.S. slowdown would risk ceding ground to China, which he says is unlikely to join any global pause.
Big Technology’s Alex Kantrowitz and Margins’ Ranjan Roy argue that Jacob Coxon’s viral warning about AI-driven human extinction has outpaced the technical case offered publicly. They do not dismiss longer-term danger, but say policy and corporate scrutiny should focus on concrete routes to harm—access controls, compute, credentials, data use, cybersecurity and shutdown mechanisms—rather than unsupported probability estimates. They also differ on whether the extinction narrative strengthens frontier labs commercially or creates regulatory and infrastructure risks for companies such as Anthropic.
Google DeepMind’s Philipp Schmid argues that agent builders should stop encoding anticipated workflows in custom functions and instead give capable models a managed sandbox, general-purpose tools, clear instructions and tightly defined access. In his three versions of a GitHub pull-request reviewer, Python loops, tool schemas and registered functions give way to an `AGENTS.md` file, skill files and a remote environment that handles execution, state and credential injection. The work that remains with teams, he says, is setting behavior, permissions and evaluation—not maintaining orchestration that improving models can replace.
Richard Socher, founder of Recursive, argues that the practical route to recursive self-improvement begins not with autonomous scientific discovery but with AI systems that can improve AI research in tightly measured environments. Recursive says its agents have surpassed prior results in small-model training and GPU-kernel optimization, but Socher’s broader claim depends on a harder condition: benchmarks, rewards and tool harnesses must reward genuine progress rather than loopholes. He contends that this capability should ultimately accelerate work across science and engineering, while regulation should target harmful applications rather than general-purpose intelligence.
Brendan Rappazzo, a machine-learning researcher at Morgan Stanley speaking about an independent open-source project, argues that people can often judge a concrete moral case but cannot anticipate every case their stated principles must cover. His tool, Loophole, translates plain-language values into a formal code, then uses adversarial agents to find conduct that is immoral but permitted or moral but prohibited. A judge agent patches drafting errors where possible and sends unresolved value conflicts back to the user for a decision.
Stanford neuroscientist Andrew Huberman argues that health protocols should build resilience rather than dependence on an ideal routine. Drawing on the framework behind his forthcoming book, he presents light exposure, exercise, sleep, breathing, difficult learning and recovery as tools for regulating arousal, tolerating discomfort and acting deliberately under stress. The aim, Huberman says, is not perfect optimization but the ability to function when the preferred tools are unavailable.
Amy Cuddy argues that bullying is not simply conflict or criticism but a serial, coordinated effort to isolate a target—and that its power depends on the bystanders who laugh, stay silent or amplify the attack. Drawing on research and her experience of a public campaign against her academic work, the social psychologist says groups can interrupt bullying early through “social bravery”: visible, shared acts of support that deny an attacker the audience and validation they need.
Simon Godwin, artistic director of Shakespeare Theatre Company, sees Jonathan Demme’s *Stop Making Sense* as a lesson in making performance visible: a lone performer with a guitar and boombox gradually becomes part of a fully populated, diverse ensemble. For Godwin, that public assembly makes the audience a participant in the creation of a shared world and raises a broader question about America’s capacity to begin with little and make room for greater possibility.
OpenAI president Greg Brockman argues that the arrival of models such as Astra marks an “AGI era,” not because they are uniformly reliable but because they can use computers and pursue work coherently over long periods. He says that shift makes safety, alignment and cybersecurity operational constraints rather than after-the-fact safeguards: organizations with frontier access should use it now to find and fix vulnerabilities before similar capabilities spread more widely.
David Rubenstein argues that Carlyle’s rise from a $5 million, deal-by-deal operation in 1987 to a firm managing $500 billion was built less on one investor’s foresight than on institutional compounding: recruiting specialists, raising capital, using Washington connections to open doors, and turning successful exits into proof for the next fund. Rubenstein, who came to private equity after careers in government and law, says his role was fundraising and building the organization while his partners led investment decisions. Even at Carlyle’s scale, he says, the entrepreneur’s job remains to assume that something can go wrong.
Psychologist Adam Mastroianni argues that incentives explain why people face pressure to do harmful things but cannot excuse conduct they know is wrong. In an EconTalk conversation with Russ Roberts, he calls surrendering moral judgment to money, status or institutional approval “loser behavior,” and contends that better systems still require people willing to refuse their terms. Their discussion tests that claim against everyday compromises, career choices and the institutions that reward them.
Professor Jiang Xueqin argues that modern technology and prosperity rest on a dollar-centred trading system whose specialized supply chains depend on American military power, functioning shipping routes and confidence in financial institutions. In his account, escalating conflict around Iran could disrupt insurable trade, cutting energy and industrial inputs to major economies and setting off a chain of scarcity, migration, political polarization and conflict. His most extreme warning is that a future attacker could seek 9/11-scale psychological shock by targeting financial records and payment infrastructure rather than physical landmarks.
Acquired’s Ben Gilbert and David Rosenthal argue that Home Depot’s enduring advantage was not the warehouse format alone, but a system that combined low prices and broad inventory with knowledgeable store associates who helped customers complete projects. They trace how Bernie Marcus and Arthur Blank built that model after their firing from Handy Dan, how Bob Nardelli’s operational overhaul weakened its service culture, and how Frank Blake restored growth by making stores the local nodes of a wider fulfillment network.
Leadership expert Chris Wheatley argues that the person receiving hard feedback—not only the person giving it—determines whether candor is safe enough to recur. Drawing on a colleague’s blunt critique of his own defensive, condescending tendencies, he sets out a four-step “TACT” method: thank the speaker, acknowledge their experience without necessarily agreeing, commit to a next step, and follow up with thanks. The aim, he says, is to reduce the personal risk of speaking honestly across relationships and hierarchies.