first-hand human knowledge
We read the podcasts, essays and interviews — and hand you the arguments: who claims what, who rebuts, and the original voice one click away.
The article traces how Ethan Mollick's guide to AI has shifted from recommending chat-based models a year ago to now emphasizing agentic systems that can do the equivalent of hours of human work autonomously.
Levin frames his central research question as how embodied minds arise in the physical world, which he unpacks through three perspectives: recognizing minds (third person), controlling/interacting with them (second person), and the first-person experience of having an inner perspective.
Demis Hassabis proposes that natural systems shaped by evolutionary processes have structure that can be learned by neural networks, making computationally intractable problems like protein folding solvable by classical systems.
Friedberg describes how Ohalo's genomics work — designing genetic constructs, predicting phenotypes — was getting restricted by Anthropic's safety filters, forcing the company to adopt Chinese open-source models instead, which are superior to American open-source alternatives.
Priscilla Chan traces the evolution from early single-cell sequencing funding to the Human Cell Atlas to the Cell by Gene annotation tool, and describes how the arrival of LLMs that can make sense of massive data finally unlocked the possibility of systematically understanding how living systems work.
The main news: OpenAI reports its next major model, an internal version of Astra, solved ten mathematical problems that had seen no progress for at least a decade, each at a claimed cost of under $2,000 at GPT-5.6 Sol token prices.
Jumper explains the decades-old bottleneck of experimentally determining protein structure and how AlphaFold's deep learning system compressed that into minutes, scaling to 200 million predicted structures.
Alex describes how the principal odor map's embedding space self-organized into perceptually meaningful, nested neighborhoods (e.g., jasmine/rose/violet inside 'floral') purely from data, mirroring biological reality.
Jumper argues AlphaFold 2 is a counterexample to the 'bitter lesson' — success came from deliberately encoding biological and geometric domain knowledge into the architecture, because protein data (unlike text) is finite.
Pushing back on the idea that AlphaFold's iterative refinement process amounts to 'understanding,' Jordan argues that anthropomorphizing AI with words like understanding and intelligence distracts from the real engineering questions.
Shapiro directly attacks Trump's fiscal policy, arguing that giving tax cuts to those who don't need them while cutting Medicaid will cause 500,000 Pennsylvanians to lose healthcare, shutter 26 rural hospitals, and increase costs for everyone else through higher private insurance premiums.
Terence Tao explains that the most interesting mathematical problems are not the impossibly hard ones, but those just on the boundary where existing techniques can do most of the work but the final step remains elusive.
OpenAI's Health in ChatGPT rollout allows US users to connect Apple Health and supported medical records, with additional encryption, a commitment not to train foundation models or target ads, and substantial physician review effort. The text argues this is less about a new model and more about a new high-trust application layer on top of existing model capability.
In the Science Corner, Friedberg describes a new Calico/Revel Pharma enzyme that clears the glycation byproduct CML from the extracellular matrix, which in a skin sample test reversed apparent skin age from over-70 to 31 years old.
Kaldellis argues historians should define this society by its long endogenous periods of internal stabilization and regrowth rather than by its brief, externally-caused crises — every crisis except the last was followed by recovery.
In an off-the-cuff whiteboard sketch, Reich proposes that Neanderthals aren't simply a separate archaic lineage modern humans later interbred with, but are descendants of a Middle-Stone-Age-inventing population related to modern humans that expanded into Europe and got genetically swamped by local archaics — while the same expansion into Africa, diluted less by more divergent archaic Africans, produced modern humans.
Reich's team found that selection intensity on the genetic predictor of IQ-test performance spiked sharply between 4,000 and 2,000 years ago, pushing the trait up by roughly a standard deviation, while the last 2,000 years show essentially no selection on it at all.
Weil recounts a personal anecdote where ChatGPT interpreted a confusing medical letter about his son's biopsy result, providing instant reassurance during a 72-hour wait to reach the doctor.
Eric Nguyen opens the episode by framing the core tension: AI's ability to generate and manipulate life offers transformative potential for biology, but also introduces serious responsibility and risk.
Terence Tao, neither dismissive of AI nor fearful of it, sees it as the catalyst for 'big mathematics' — a future of large-scale, decentralized human-machine collaboration in which humans take the creative parts and AI does most of the technical grunt work.