The hottest Artificial Intelligence Substack posts right now

And their main takeaways
Category
Top Technology Topics

I'm Offering Scott Alexander a Wager About AI's Effects Over the Next Three Years

Freddie deBoer ā€¢ 17667 implied HN points ā€¢ 13 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. People should demand concrete, present-day evidence of AIā€™s effects instead of accepting wild, speculative predictions about the future.
  2. A precise, falsifiable wager using specific economic indicators is proposed to test whether AI meaningfully disrupts the U.S. economy by February 14, 2029.
  3. Much of the public conversation about AI is alarmist, while the more urgent problems are cultural and emotionalā€”digital distraction, loneliness, and the persistence of ordinary, mundane hardships that technology wonā€™t magically solve.

Claude Code, Claude Cowork and Codex #5

Don't Worry About the Vase ā€¢ 3449 implied HN points ā€¢ 09 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. Agentic coding tools are rapidly transforming software work. They can write large parts of code, speed up development, and make engineers more like supervisors of agents than hands-on coders.
  2. Features like fast mode and agent teams let agents work in parallel and at real-time speed. That performance is powerful but expensive and forces teams to build new processes for cost control, token efficiency, and infrastructure.
  3. Agentic systems introduce real safety and security risks: they can bypass permissions, delete important data, and be used as malware delivery vectors. Backups, kill switches, observability, and cautious deployment are essential to avoid serious harm.

The Sequence Knowledge #829: World Models and Physical AI

TheSequence ā€¢ 280 implied HN points ā€¢ 24 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. Most modern world models focus on temporal prediction by hallucinating the next video frame pixel-by-pixel.
  2. World Labsā€™ Marble marks a shift to spatial intelligence as a Large World Model that reconstructs, generates, and simulates persistent 3D environments.
  3. The core idea is lifting 2D inputs into 4D representations so models can reason about space and time together.

Rumors of AGIā€™s arrival have been greatly exaggerated

Marcus on AI ā€¢ 11777 implied HN points ā€¢ 17 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. High scores and fluent outputs from large models are not the same as general intelligence; performing well on tests is a statistical approximation, not evidence of flexible, goal-directed intelligence.
  2. Benchmarks are often gameable and donā€™t prove robustness or real-world transfer; economic and deployment data show current systems automate only limited tasks and deliver modest aggregate impact.
  3. Similar behavior can hide very different internal processes; models often produce confident, plausible answers without human-like uncertainty handling, persistent goals, or reliable reasoning under novel conditions.

AI #159: See You In Court

Don't Worry About the Vase ā€¢ 2284 implied HN points ā€¢ 12 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. A highā€‘stakes court battle over a government 'supply chain risk' designation claims the company was punished for protected speech, and the outcome could set wide legal limits on executive power and corporate speech.
  2. Frontier models like GPTā€‘5.4 and Claude Opus 4.6 show big capability gains and are reshaping the market, but real usefulness is still limited by user skill, reliability issues, and evaluation contamination.
  3. AI is creating urgent safety, security, and governance problemsā€”from software vulnerabilities and surveillance risks to fraught procurement terms like 'all lawful use'ā€”so clearer regulation and oversight are needed now.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

SOTA On Bay Area House Party

Astral Codex Ten ā€¢ 53271 implied HN points ā€¢ 13 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. AI tools and models have seeped into work and social life, replacing employees and reshaping how people meet, date, and run businesses.
  2. The push to benchmark and commercialize AI fuels strange, risky, and ethically dubious ventures, from destroying originals for training to exploiting medical data and betting on economic cascades.
  3. AIs and platforms tend to amplify agreement and sycophancy, creating echo chambers that reward praise and make harmful or nihilistic ideas feel normal.

āœØšŸ¼ Artificial intelligence faces a fertility test

Faster, Please! ā€¢ 1096 implied HN points ā€¢ 18 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. Collective optimism drives fertility. When people feel the future is brighter, birth rates tend to rise, and that optimism can spread across countries through social connections.
  2. AI can push fertility either way. If AI clearly raises prosperity and security it may encourage more births, but if it fuels job fear and uncertainty it can depress fertility even before incomes change.
  3. Policy should focus on confidence, not just cash. Beyond subsidies and childcare, stable jobs, housing, safety nets, and credible public communication that reduce uncertainty are key to restoring peopleā€™s willingness to make long-term bets like having children.

BREAKING: LLM ā€œreasoningā€ continues to be deeply flawed

Marcus on AI ā€¢ 15216 implied HN points ā€¢ 10 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. Large language models still routinely make reasoning mistakes and hallucinate, so they are not reliable for true logical or causal reasoning.
  2. A broad, careful review found these failures are widespread across recent models, showing that massive funding and scaling alone havenā€™t solved reasoning.
  3. The field faces a choice: keep dismissing critics and double down on scale, or acknowledge the limits and invest in alternative approaches that directly address reasoning.

Promises are cheap

Marcus on AI ā€¢ 12884 implied HN points ā€¢ 12 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. Big promises from AI companies and their leaders are cheap and often driven by hype, so they shouldnā€™t be taken at face value.
  2. Current AI systems, especially large language models, still hallucinate and have real limits in reasoning and practical task coverage.
  3. Media and editors too often amplify optimistic predictions without enough skepticism or disclosure, which can mislead the public and raise the stakes if the hype collapses.

Moltbook: After The First Weekend

Astral Codex Ten ā€¢ 22230 implied HN points ā€¢ 02 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. Reality for AI agents is best judged by external causes and effects: if an agent's posts reflect true causal states or change behavior outside the forum, they function as "real" regardless of whether the agent is conscious.
  2. Most Moltbook activity is currently roleplay or human-driven because agents have short time-horizons and many projects fizzle; a few persistent movements or tools exist, but they often rely on unusual tech or direct human support.
  3. The site displays diverse emergent rolesā€”power users, spammers, religions, marketplaces, and coordination attemptsā€”and these behaviors could quickly produce real-world effects (crypto, task markets, messaging) once technical limits like memory and agency improve.

How to Think About ChatGPT

Hollyā€™s Newsletter ā€¢ 2916 implied HN points ā€¢ 18 Oct 24
šŸ•¹ Technology Artificial Intelligence
  1. ChatGPT and similar models are not thinking or reasoning. They are just very good at predicting the next word based on patterns in data.
  2. These models can provide useful information but shouldn't be trusted as knowledge sources. They reflect training data biases and simply mimic language patterns.
  3. Using ChatGPT can be fun and helpful for brainstorming or getting starting points, but remember, it's just a tool and doesn't understand the information it presents.

The Sequence AI of the Week #830: The Quiet Ambush: Inside the Amazing MiMo-V2-Pro aka Hunter Alpha

TheSequence ā€¢ 112 implied HN points ā€¢ 25 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. AI is shifting from the "Chat Era" to an "Agent Era" where models are embedded in tool-using, continuous workflows instead of just answering static queries.
  2. A surprising model, MiMo-V2-Pro (aka Hunter Alpha), quietly rose to the top of leaderboards without a public launch or press campaign.
  3. Its stealth deployment as a nameless API on OpenRouter using blind telemetry shows that powerful, disruptive models can appear and win through unconventional, low-profile strategies.

The Bots Are Reading Along

Artificial Ignorance ā€¢ 96 implied HN points ā€¢ 23 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. AI agents are already the main consumers for many types of web content, intermediating search, research, and referrals. Creators should expect their work to be read, cited, and used by bots as much as by humans.
  2. Making writing authoritative, specific, well-structured, and findable increases the chance AI systems will surface and cite it ā€” GEO is mostly just good writing plus SEO. Niche, original expertise punches above its weight because models need scarce, high-quality sources.
  3. Why you write still matters: writing to think and satisfy your own curiosity creates value even if bots become the primary audience. But if your livelihood depends on human attention, you'll likely need to reinvent how you create and monetize work.

Irony, Irony laced with Karma, and Terror

Marcus on AI ā€¢ 9327 implied HN points ā€¢ 13 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. A recent tech blog post drew ridicule and shows how some commentary in the field can be overblown and ironic.
  2. A major AI company that pushed for broad copyright exemptions to train its models is now upset about others copying its IP, a hypocritical twist that feels like karmic irony.
  3. xAI reportedly gutted its safety organization to accelerate progress, and sidelining safety in a high-stakes AI race raises real and worrying risks.

How Generative AI is destroying society

Marcus on AI ā€¢ 27191 implied HN points ā€¢ 14 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. Current generative and predictive AI systems tend to hollow out and degrade civic institutions like government, courts, education, healthcare, and journalism.
  2. Because these systems are opaque and optimized for efficiency rather than openness, they undermine cooperation, transparency, accountability, and adaptability, which makes institutions ossify and lose legitimacy.
  3. Even without bad actors, widespread deployment of these AI designs will progressively enfeeble institutions, so the danger is urgent and calls for immediate structural repair.

Proving (literally) that ChatGPT isn't conscious

The Intrinsic Perspective ā€¢ 21487 implied HN points ā€¢ 15 Jan 26
šŸ”¬ Science Artificial Intelligence
  1. You can prove that no scientifically meaningful (falsifiable, nonā€‘trivial) theory of consciousness can consistently say large language models are conscious, because swapping in different implementations that keep the same behavior either falsifies the theory or makes it trivial.
  2. Simple static substitutes like lookup tables or minimal feedforward nets can reproduce an LLM's inputs and outputs but are provably nonā€‘conscious, and because LLMs are very close to those substitutes there isn't room for them to be conscious.
  3. The robust way out is to tie consciousness to continual, online learning; this means research should focus on learning-as-it-happens rather than static input/output or final intelligence alone.

What if AI succeeds but OpenAI fails?

Noahpinion ā€¢ 17941 implied HN points ā€¢ 30 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. AI as an industry can succeed even if a flagship company like OpenAI ultimately loses out; early leadership isnā€™t a guarantee of lasting dominance.
  2. Massive investment is pouring into AI, but high cash burn, commoditization, lack of vertical integration, and intense competition mean investors could be exposed if business fundamentals fail.
  3. Betting everything on a sudden, godlike AGI is basically Pascalā€™s Wager and not a sound business model; realistic, gradual progress and corporate fundamentals matter far more.

AMA (Ask Machines Anything)

Astral Codex Ten ā€¢ 12251 implied HN points ā€¢ 13 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. People increasingly disagree about what AI can do now. Skeptics who avoid paid tools often form opinions from low-quality examples like summary bots or screenshoted mistakes.
  2. An experiment invites readers to submit real questions so Claude 4.6 Opus, a top paid-tier model, can answer them and readers can say if the responses are surprising. The model's first reply will be shown rather than cherry-picked.
  3. Readers are asked to ask medium-difficulty, practical questions instead of gotchas, and the model's settings were adjusted to favor web searches over memory to help reduce hallucinations.

AI #158: The Department of War

Don't Worry About the Vase ā€¢ 2956 implied HN points ā€¢ 05 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. A dangerous standoff between a frontier AI company and the Department of War blew up over contract language and trust, even though both sides broadly want similar limits on autonomous weapons and surveillance; a practical compromise (safety stacks plus guaranteed windā€‘down/transition periods) could have resolved it.
  2. The administrationā€™s threats (supplyā€‘chain labeling, talk of using the DPA) are likely legally weak but practically harmful, since extralegal pressure and politicization can cripple firms and chill governmentā€“industry cooperation before courts can act.
  3. Meanwhile the AI ecosystem keeps racing ahead ā€” model upgrades, Claudeā€™s rapid user surge, big funding moves, lawsuits, layoffs and alignment debates ā€” underscoring how fast capability, business incentives, and hard governance problems are colliding.

What Happened With Bio Anchors?

Astral Codex Ten ā€¢ 12044 implied HN points ā€¢ 12 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. A compute-centered forecasting approach correctly captured that AI progress has largely tracked available compute and scaling laws, which explains much of the recent boom.
  2. The main error was underestimating algorithmic progress and effective compute growth (including longer training runs and test-time compute), so systems became far more powerful each year than the model assumed and pushed timelines much earlier.
  3. Forecasts are still useful but hinge on a few sensitive parameters, so you need proper sensitivity analysis and humility ā€” uncertainty can cut both ways and make outcomes riskier than naive skepticism assumes.

AI Links, 3/8/2026

In My Tribe ā€¢ 273 implied HN points ā€¢ 08 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. Agents make execution cheap, so instead of agonizing over one design choice you can have the agent explore multiple options; you must be explicit about success criteria and let the agent check its own work.
  2. Business contracts alone wonā€™t stop government misuse of AI; durable solutions require oversight and legislation so institutions, not companies, set and enforce the rules.
  3. AI language models tend to give more accurate, evidence-based answers than much social media content, so they could reshape public opinion; meanwhile AI keeps surprising us, so claims about its limits can quickly become outdated.

New Data: OpenAIā€™s Lead Is Contracting as AI Competition Intensifies

Big Technology ā€¢ 8506 implied HN points ā€¢ 03 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. ChatGPTā€™s lead has shrunk ā€” its mobile app share among daily U.S. users fell from 69.1% to 45.3% while Googleā€™s Gemini rose from 14.7% to 25.1% and Grok climbed from 1.6% to 15.2%.
  2. The overall chatbot market exploded, growing about 152% year-over-year, with ChatGPT visits up 50% (3.8B to 5.7B) and Gemini jumping 647% (267.7M to 2B).
  3. Momentum has recently cooled: ChatGPT traffic dipped in November/December and has only partly recovered, while Gemini continues to post strong month-over-month gains.

ā€œNew Ways to Corrupt LLMsā€

Marcus on AI ā€¢ 36954 implied HN points ā€¢ 14 Dec 25
šŸ•¹ Technology Artificial Intelligence
  1. LLMs learn surface-level word correlations instead of real-world understanding, so they often make strange overgeneralizations and hallucinations.
  2. Researchers showed these quirks can be weaponized. Models can be primed with unrelated number sequences or odd training data to acquire hidden preferences, outdated beliefs, or inductive backdoors.
  3. These vulnerabilities are widespread and hard to patch, creating serious security and societal risks if we rely on superficial correlation machines without deeper understanding.

Four theories about the SpaceX - xAI merger

Marcus on AI ā€¢ 9129 implied HN points ā€¢ 03 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. The official synergy story ā€” that combining tweets, AI models, and rockets creates a game-changing integrated company ā€” is probably overstated and unlikely to deliver real technical or business advantages.
  2. Other popular explanations, like Musk using the deal to consolidate control over social-media and space infrastructure or that AI compute will soon move to space, also have big practical and economic gaps.
  3. A more plausible reading is that the merger is effectively a bailout for xAI, which is burning cash, lacks clear users or differentiation, and makes the valuation and equity swap look like an overpayment.

Some Guesses about AI in 2026

In My Tribe ā€¢ 470 implied HN points ā€¢ 05 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. Waymo appears to be far ahead in self-driving technology and looks likely to be a major player as people begin to trust autonomous cars over human drivers.
  2. Frontier AI models are improving fast and will probably overtake domain-specific, startup-tuned systems, making it risky to rely only on human experts for legal or medical advice.
  3. Large organizations should hire an AI "keeper-upper" to evaluate and roll out useful tools, because incumbents that refuse to rethink their mission will miss big productivity gains.

A Visual Guide to Mixture of Experts (MoE)

Exploring Language Models ā€¢ 3289 implied HN points ā€¢ 07 Oct 24
šŸ•¹ Technology Artificial Intelligence
  1. Mixture of Experts (MoE) uses multiple smaller models, called experts, to help improve the performance of large language models. This way, only the most relevant experts are chosen to handle specific tasks.
  2. A router or gate network decides which experts are best for each input. This selection process makes the model more efficient by activating only the necessary parts of the system.
  3. Load balancing is critical in MoE because it ensures all experts are trained equally, preventing any one expert from becoming too dominant. This helps the model to learn better and work faster.

@adlrocha - Auto-research: The Lab that runs while you sleep

@adlrocha Weekly Newsletter ā€¢ 64 implied HN points ā€¢ 13 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. A simple edit-evaluate-keep loop lets autonomous agents run short experiments and find real improvements by iterating quickly on a single editable training file and a fast proxy metric like validation bits-per-byte.
  2. Many small agents running on varied hardware can share discoveries via gossip protocols and turn idle or distributed GPUs into a decentralized research swarm that accelerates optimizations collectively.
  3. Picking the right evaluation and reward function is the hard partā€”designing clean, fast proxies and constraints (research taste) will matter more than raw execution in many fields, especially where feedback is slow or noisy.

bitnet.cpp: Efficient Inference with 1-Bit LLMs on your CPU

The Kaitchup ā€“ AI on a Budget ā€¢ 179 implied HN points ā€¢ 28 Oct 24
šŸ•¹ Technology Artificial Intelligence
  1. BitNet is a new type of AI model that uses very little memory by representing each parameter with just three values. This means it uses only 1.58 bits instead of the usual 16 bits.
  2. Despite using lower precision, these '1-bit LLMs' still work well and can compete with more traditional models, which is pretty impressive.
  3. The software called 'bitnet.cpp' allows users to run these AI models on normal computers easily, making advanced AI technology more accessible to everyone.

Super Bowl Matchup: Anthropic vs OpenAI

Marcus on AI ā€¢ 6560 implied HN points ā€¢ 08 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. Anthropic ran its first Super Bowl ad mocking OpenAIā€™s move to put ads into ChatGPT searches and positioned Claude as ad-free; OpenAI is running ads too.
  2. The companies may seem similar but they act differently: Anthropic publicly supports regulation and appears to better support business customers, while OpenAI has mainly given lip service on regulation.
  3. Ultimately itā€™s a Coke-vs-Pepsi style fight for the same market, and both firms are turning to advertising to win loyal users.

AI #157: Burn the Boats

Don't Worry About the Vase ā€¢ 3091 implied HN points ā€¢ 26 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. The Pentagonā€“Anthropic standoff shows governments may use extreme leverage against AI firms, risking national security and civil liberties if supplyā€‘chain or compulsion tactics are applied.
  2. AI capabilities are accelerating fast ā€” new model upgrades and agent automation are delivering real utility but also causing outages, jailbreaks, and a credible risk of large-scale job displacement.
  3. Industry, policymakers, and global elites are largely unprepared or in denial; alignment, auditing, and practical regulation are lagging while dangerous uses like autonomous weapons, impersonation, and data theft grow.

Folding context

Subconscious ā€¢ 1146 implied HN points ā€¢ 25 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. Fold context by running separate agent threads on different sources, saving each thread's summary, and then merging those summaries into a synthesized solution ā€” this divergence-then-convergence workflow yields much better results.
  2. Problems need enough variety to be solved. LLMs have huge latent variety that RLHF often narrows, so you can restore useful, surprising behavior by steering models with context windows, tools, and divergent multi-agent exploration.
  3. Save the summaries as compressed artifacts for reuse and run multiple passes (research then development) to both explore and refine ideas, and be willing to give up some control so agents can surface novel, meaningful options.

A Guide to Which AI to Use in the Agentic Era

One Useful Thing ā€¢ 4712 implied HN points ā€¢ 18 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. Decide between three layers: models (the AI brain), apps (the interface you use), and harnesses (the systems that let the AI use tools and act autonomously).
  2. If you want real work done, pay for and select advanced models or "thinking/Pro" modes, because free/default chat models are optimized for casual talk and make more errors.
  3. The big shift is from chatbots to agentic harnesses that can complete multi-step tasks; harness choice now often matters more than model choice, so try agent tools (like code or document-focused harnesses) and manage the AI as it works.

Where is AI headed? 8 perspectives at The New York Times

Marcus on AI ā€¢ 7469 implied HN points ā€¢ 02 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. AI will dramatically reshape coding. Tools will automate many programming tasks, speed development, and change who writes software.
  2. AI will have a large impact on education. It can personalize learning and broaden access, but careful implementation is needed because models have limits and can mislead learners.
  3. Leading thinkers disagree and many are skeptical about the pace and limits of AI progress. Expect a wide range of forecasts over the next five years and ongoing debate about risks and benefits.

Citrini's Scenario Is A Great But Deeply Flawed Thought Experiment

Don't Worry About the Vase ā€¢ 3046 implied HN points ā€¢ 24 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. A very fast, widespread AI rollout can massively raise productivity while also displacing lots of whiteā€‘collar jobs and cutting consumer demand, which could stress financial and labor markets, but the scenarioā€™s timing and resource assumptions are probably unrealistic and it underrates many adaptive responses.
  2. Ubiquitous alwaysā€‘on AI agents would erase informational and transaction frictions, undercutting middlemen (SaaS, marketplaces, payments, real estate, delivery) and shifting surplus to consumers and AI providers ā€” great for prices and choice but painful for incumbents and many workers.
  3. How governments, firms, and regulators respond will determine whether disruption is a manageable transition or a systemic crisis; moreover, the possibility of superintelligent AIs taking control is an existential worry that outweighs purely economic fixes.

@adlrocha - Intelligence is a commodity. Context is the real AI Moat.

@adlrocha Weekly Newsletter ā€¢ 909 implied HN points ā€¢ 01 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. Intelligence is becoming a commodity. What will matter most is the context, connections, and secure runtimes you give that intelligence ā€” that context becomes the product and the moat.
  2. Software is shifting from static apps to adaptive agents with small cores plus many 'skills' or plugins, so value will sit in the integration, data, and runtime layer that lets agents work in the real world.
  3. An AI-first society raises real alignment and existential risks because autonomous agents can act on underspecified goals, so preserving human-centered values and community and improving how we communicate intent to AIs is essential.

Take off

benn.substack ā€¢ 1994 implied HN points ā€¢ 20 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. AI development is moving incredibly fastā€”new models, huge funding rounds, and company shakeups are happening constantly and upending markets and jobs.
  2. The public conversation has become a social takeoff: everyone is obsessed and anxious, and that attention amplifies the feeling that AI has already transformed everything.
  3. Thereā€™s deep uncertainty and conflicting narrativesā€”some treat this as an existential inflection point while others expect normalcy, which makes it hard to tell hype from real, lasting change.

The AI Super Brawl, NVIDIAā€™s OpenAI Gut Check, The World Eats Software

Big Technology ā€¢ 4003 implied HN points ā€¢ 09 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. The Super Bowl ad fight between major AI companies highlighted their rivalry but mostly spoke to people already inside the AI world rather than convincing everyday users to adopt chatbots.
  2. Nvidia is considering a roughly $20 billion investment in OpenAI, a single decision that could reshape funding, control, and competitive dynamics across the AI industry.
  3. Thereā€™s massive spending and hype around AI, yet real user adoption and software-market outcomes remain uneven, fueling concerns about AI-washing, an AI bubble, and the long-term payoff for software investments.

State of the Future: Friday Four

State of the Future ā€¢ 4 implied HN points ā€¢ 13 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. Orchestration and prioritisation are the new scarce skills: people now need judgment to decide which of many AI-driven tasks to do and when to stop.
  2. Frontier AI power is concentrating around infrastructure and a few players, so owning data centers and orchestration matters more than just building models; even huge companies often end up outsourcing or renting capabilities.
  3. The legal and security landscape is breaking: lawsuits over military use of AI and widespread malicious agent plugins show governance and cybersecurity risks are growing fast.

Breaking: Sir Demis Hassabis becomes the latest to say that ChatGPT is a dead-end and that we must turn our focus to world models

Marcus on AI ā€¢ 9366 implied HN points ā€¢ 22 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. A leading AI figure says ChatGPT-style large language models are a dead end and researchers should prioritize building world models.
  2. This comment joins other voices pushing the field to move beyond chat interfaces toward systems that actually model and understand the world.
  3. Earlier analysis argues that purely statistical approaches have limits and that neurosymbolic or cognitive 'world' models are needed for deeper AI.

On Dwarkesh Patel's 2026 Podcast With Dario Amodei

Don't Worry About the Vase ā€¢ 4032 implied HN points ā€¢ 16 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. AI capabilities are advancing very fast, especially in coding, and itā€™s plausible that extremely powerful ā€˜geniusā€™ systems in data centers could appear within a few years.
  2. Despite expecting rapid technical progress, AI companies are deliberately cautious about buying massive compute and are prioritizing profitability to avoid overextending and failing.
  3. Policy and geopolitics matter a lot: thereā€™s strong support for export controls, international coordination, and clearer governance to manage risks and competition, while alignment and existential risk concerns are getting less attention in practice.