Interview, Explained
In-depth breakdowns of tech interviews — section-by-section analysis with diagrams, quotes, and insights.
~92 minEric Nguyen on why the model that writes DNA must also guard it
Radical Numerics' new Omni model shows DNA language models can now beat specialized tools at predicting disease-causing mutations, and Eric Nguyen argues the same generative models must be turned into sequence-level biosecurity defenses.
~60 minNeal Mohan on Why the Audience, Not AI, Runs YouTube
YouTube's CEO argues AI is the stage, not the star: the algorithm is really the audience voting, the line worth policing is quality over slop, and the moat is human storytelling that AI can't fake.
~43 minRenen Hallak on why your IP is becoming model weights
VAST Data's Renen Hallak argues that as company value shifts from people and data to model weights, trust has to move from promises to math: encrypt the weights, run inference on-prem, and confidential computing can unlock the enterprise data AI has not touched yet.
~10 minFei-Fei Li on Why World Models Generate Worlds, Not Words
Fei-Fei Li argues the next AI substrate is spatial intelligence — world models that generate navigable 3D worlds instead of words — and that the harder questions of safety, benchmarking, and governance are human problems to be solved at the ecosystem level, not left to any single lab or to fatalism.
~6 minNeil Chilson on why a safety waiver risks a durable AI cartel
A former FTC chief technologist argues frontier labs do not need an antitrust waiver to build AI safely, and warns the exemption they want is exactly what turns a fragile cartel into a durable, government-backed one.
~9 minGreg Allen on why AI capability is outrunning our safety tools
Greg Allen argues that safety capability is falling behind raw AI capability, and that self-replicating models, open weights, and the US-China race all widen the gap that reciprocal safety commitments would have to close.
~9 minChase Lochmiller on why building AI starts with the electron, not the chip
Crusoe's CEO argues that building AI is as much a physical job as a software one, so his company runs the whole chain from generating power to serving a model's output, and now has to win the local politics of building data centers before the backlash sets in.
~38 minStefano Ermon on why diffusion models are built for the GPU
Stefano Ermon, a pioneer of diffusion models and CEO of Inception, argues the next AI edge may come as much from fitting the model to the GPU as from making it smarter: diffusion generates tokens in parallel, so inference stops being a memory-bound crawl and starts using the chip the way training does.
~67 minAli Ghodsi on why the AI bottleneck is context, not intelligence
Databricks CEO Ali Ghodsi argues the existential-risk panic is overblown - near-term p(doom) is close to zero - while the real work is unglamorous engineering: securing systems against fast-moving cyber threats and giving already-capable models the organizational context they lack.
~80 minNoam Brown on why the model, not the swarm, did the hard part
OpenAI's Noam Brown explains how many agents scale test-time compute in parallel, why the underlying model deserves the credit for cracking a Millennium Prize Problem, and why aligning these systems is the unsolved problem that worries him most.
~39 minGorkem Yurtseven and Batuhan Taskaya on generating video faster than it plays
fal post-trained MiniMax's open-source H3 video model into H3 Max, combining model-side post-training with kernel and systems work to generate a five-second clip in about 1.5 seconds. Generating video faster than it plays turns it from a batch render you wait on into a live medium you can direct.
~11 minMarc Benioff on why AI can't become social media 2.0
Marc Benioff argues that AI's real power shows up only when its probabilistic guessing is paired with the deterministic controls of enterprise software, and that the companies and executives building these systems must take responsibility for what they ship so AI does not repeat social media's harms.