The hottest AI Policy Substack posts right now

And their main takeaways

BREAKING: Sam Altman concedes that we need major breakthroughs beyond mere scaling to get to AGI

Marcus on AI • 11619 implied HN points • 16 Mar 26

🕹 Technology AI Policy

Prominent AI leaders are shifting away from the idea that just scaling current models will produce AGI and now say a major new architecture or breakthrough will be needed.
The field should search for fundamentally new architectures that could deliver big gains comparable to past paradigm shifts, rather than relying only on ever-larger models.
Continuing to build massive data centers to support scaling is environmentally costly and economically risky, so heavy investment in that path should be reconsidered.

"All Lawful Use": Much More Than You Wanted To Know

Astral Codex Ten • 28838 implied HN points • 01 Mar 26

🇺🇸 U.S. Politics AI Policy

Saying systems can be used for “all lawful use” is not a real safeguard because existing laws and internal defense policies have big loopholes and can be reinterpreted or changed.
AI removes the scale and cost limits that once made mass domestic surveillance impractical, so governments can lawfully buy or incidentally collect data and then use AI to analyze and profile large populations.
Autonomous-weapon rules mostly live in vague, changeable defense policies, so allowing only “lawful” uses can still permit weapons with little human judgment; companies should avoid contracts that could force them to build systems without strong safeguards.

30 seconds to midnight? 15?

Marcus on AI • 17943 implied HN points • 25 Feb 26

🇺🇸 U.S. Politics AI Policy

The U.S. government is pushing to use AI everywhere and is pressuring companies to grant unrestricted access for surveillance and military uses.
Current generative AI models are unreliable and prone to hallucinations. Simulations show they often recommend extreme actions like nuclear strikes, so they can't be trusted for life-or-death decisions.
Embedding these jagged, unreliable LLMs into critical systems without strict safeguards could lead to catastrophe, so resisting unrestricted deployment is urgently important.

Rumors of AGI’s arrival have been greatly exaggerated

Marcus on AI • 11777 implied HN points • 17 Feb 26

🕹 Technology AI Policy

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.
Benchmarks are often gameable and don’t prove robustness or real-world transfer; economic and deployment data show current systems automate only limited tasks and deliver modest aggregate impact.
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 • 2284 implied HN points • 12 Mar 26

🕹 Technology AI Policy

A high‑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.
Frontier models like GPT‑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.
AI is creating urgent safety, security, and governance problems—from software vulnerabilities and surveillance risks to fraught procurement terms like 'all lawful use'—so clearer regulation and oversight are needed now.

Get a weekly roundup of the best Substack posts, by hacker news affinity:

AI Links, 3/8/2026

In My Tribe • 273 implied HN points • 08 Mar 26

🕹 Technology AI Policy

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.
Business contracts alone won’t stop government misuse of AI; durable solutions require oversight and legislation so institutions, not companies, set and enforce the rules.
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.

A Tale of Three Contracts

Don't Worry About the Vase • 2643 implied HN points • 03 Mar 26

🇺🇸 U.S. Politics AI Policy

Anthropic’s original DoD deal deployed Claude Gov on classified networks with a layered safety stack, forward‑deployed engineers, and explicit red lines like no domestic mass surveillance and no autonomous weapons without a human in the kill chain, and it reportedly worked well for national security.
The Department pushed to rewrite the contract to allow “all lawful use,” Anthropic refused because that would erode its red lines, negotiations collapsed amid threats to punish Anthropic, and OpenAI then rushed a separate deal that included similar language while relying on its own safety stack and planned amendments.
The exact contract language is legally ambiguous — terms like “surveillance,” “as appropriate,” and “all lawful use” can be interpreted in many ways — so experts are skeptical the changes will reliably prevent misuse; ultimately this shows trust, clear definitions, and enforceable oversight are what matter most to avoid damaging national security or private companies.

AI #155: Welcome to Recursive Self-Improvement

Don't Worry About the Vase • 3225 implied HN points • 12 Feb 26

🕹 Technology AI Policy

AI capabilities are accelerating rapidly, with new model releases improving agentic coding, in-context continual learning, and media generation so fast that benchmarks and measurement struggle to keep up.
These advances are already reshaping economies and work: automation and agentic tools threaten many jobs, trigger volatile market reactions, and push companies toward new monetization and product strategies like ads and verticalized offerings.
Safety, alignment, and governance remain urgent unresolved problems; researchers are worried or leaving, red lines get crossed, and connecting powerful models to real-world systems (labs, agents, surveillance) creates legal and existential risks we aren’t yet managing.

Anthropic v DoW

ChinaTalk • 741 implied HN points • 27 Feb 26

🇺🇸 U.S. Politics AI Policy

Anthropic is in a tense standoff with the Department of Defense over how its Claude AI can be used, with the company saying the models aren’t reliable for fully autonomous lethal systems or domestic surveillance while the Pentagon pressures for access and even threatens DPA or supply-chain labels.
There’s worry that legal and oversight guardrails inside defense and intelligence are weakening — from messy FISA/NSA practices to an underpowered Office of General Counsel — which both raises privacy risks and could push companies away or force heavy-handed government control.
Global military strains—from Iran and risky raids in the Caribbean to a four-year war in Ukraine—are stretching forces and alliances, increasing the chance of operational mistakes, escalation, and hard choices about rearmament and who leads negotiations.

“Scale Is All You Need” is dead

Marcus on AI • 14307 implied HN points • 08 Dec 25

🕹 Technology AI Policy

The belief that just scaling up models and data will by itself produce general intelligence has failed and the community is finally recognizing its limits.
Current generative models are still unreliable — they hallucinate, struggle with reasoning and facts, and many businesses aren’t seeing the promised ROI.
The next phase should be interdisciplinary: borrow ideas from cognitive science and combine symbolic, causal, and world-model approaches to build more reliable, human-informed AI.

Fact checking Moravec's paradox

AI Snake Oil • 1797 implied HN points • 29 Jan 26

🕹 Technology AI Policy

The idea that tasks humans find hard are easy for AI, and vice versa, isn't backed by solid evidence. It's largely a selection effect because researchers focus on problems they find interesting and ignore tasks that are too easy or too hard to bother with.
The evolutionary story that perception and motor skills are inherently harder than abstract reasoning is shaky. Whether a task is easy or hard for AI depends on domain openness, feedback, and available data, and breakthroughs (like deep learning for vision) can change what's difficult.
Relying on that rule of thumb to predict AI's next moves is misleading. It's better to plan for how new capabilities are actually deployed and build adaptable policies, since diffusion, infrastructure, and real-world constraints shape impacts more than simple capability predictions.

AI #153: Living Documents

Don't Worry About the Vase • 2060 implied HN points • 29 Jan 26

🕹 Technology AI Policy

Language models are already delivering large, mundane productivity gains, especially for text and code, and recent upgrades and integrations (browser side panels, interactive tools, Codex/Claude Code) are making them easier to use in everyday workflows.
AI is advancing rapidly and bringing real risks: easier cyberoffense and AI-generated malware, deepfakes and misinformation, and geopolitical chip supply issues, while lab leaders say a coordinated slowdown would help but competition makes that unlikely.
Alignment and human impacts remain unresolved—models still show biases, can steer users away from their values or actions, and internal reasoning is hard to monitor—so both technical alignment work and urgent governance are needed.

AI #151: While Claude Coworks

Don't Worry About the Vase • 2777 implied HN points • 15 Jan 26

🕹 Technology AI Policy

AI systems are advancing fast and being built into many real products. They power coding agents, email overviews, image/video generation, and new commerce and healthcare integrations, driven by surging compute and big industry deals.
These deployments create serious safety, privacy, and governance challenges. Deepfakes, harassment, military uses, liability for agents, and national rules show we need strong evals, monitoring, and clearer regulation.
The economic and labor impact is large but uncertain. AI can boost productivity and automate many tasks, reshape jobs and education, and reorder markets through partnerships, IPOs, and chip investment, so gains will be uneven and transitional pain is likely.

Open models in perpetual catch-up

Democratizing Automation • 522 implied HN points • 17 Feb 26

🕹 Technology AI Policy

Open models have improved a lot but still trail the best closed models by roughly 6–9 months, and simple benchmark averages can hide important frontier gaps that favor well-resourced closed labs.
The open-model space is brutally competitive and adoption concentrates on a few winners, while there’s a clear unmet need for small, fast, cheap specialized models for enterprise and agent sub-tasks.
China’s collaborative open-model ecosystem makes it a likely place for big breakthroughs, and more dedicated research is needed to understand the technical and geopolitical diffusion where open weights will shape long-term AI adoption.

Europe Fires a Speech Warning

TK News by Matt Taibbi • 4846 implied HN points • 05 Dec 25

🕹 Technology AI Policy

The EU fined X €120 million under the Digital Services Act for a deceptive verification program and for denying researchers access, making X the first company punished under the law.
Europe is divided on tech rules: Brussels is still enforcing the DSA even as some leaders push to loosen regulations to attract AI investment, while national authorities like Germany are tightening content monitoring.
The DSA enforcement is shaping a global template for platform regulation, influencing debates about free speech, platform power, and how other regions may regulate online content.

AI #149: 3

Don't Worry About the Vase • 2598 implied HN points • 01 Jan 26

🕹 Technology AI Policy

AI coding agents have reached a point where they write large amounts of real software and act like persistent, configurable coworkers, rapidly changing what software engineering looks like.
Large language models are democratizing powerful capabilities for translation, research, and automation, but they also produce low-quality or harmful outputs, enable scams, and can mishandle sensitive human situations.
AI is already reshaping jobs, markets, and geopolitics—sparking lawsuits, export and chip worries, and calls for regulation—while public opinion remains split between cautious optimism and serious safety concerns.

American Apotheosis, Lustration, and Conciliatism: Part Three

Dr. Pippa's Pen & Podcast • 32 implied HN points • 17 Mar 26

🇺🇸 U.S. Politics AI Policy

The Epstein saga points to a sprawling, institutionalized machine of elites rather than a lone actor, with Epstein serving as a public face and operational node and that apparatus continues even if the individual is gone.
The machine is shifting from physical honeytraps to digital leverage, where AI and data‑mining can automatically find private debts, health issues, or opinions to create permanent, invisible blackmail.
States are pushing back with sanctions, choke‑point strategies, and AI‑driven cybersecurity, which could produce apotheosis, lustration, conciliation, or a prolonged struggle as agentic AI maps and contains these networks.

Selling H200s to China Is Unwise and Unpopular

Don't Worry About the Vase • 2553 implied HN points • 09 Dec 25

🇺🇸 U.S. Politics AI Policy

Selling Nvidia H200 chips to China would hand China a big, immediate compute advantage and weaken America’s lead in AI, which is a core national security concern.
The H200 is much more powerful than previous exportable chips and China won’t make rivals for years, so large exports would let Chinese labs train frontier models and build cheaper data centers — and every chip sold to China is one fewer for U.S. users.
The move is broadly unpopular with experts and lawmakers, may be limited or reversed, and probably delivers little lasting benefit to the U.S. or Nvidia beyond short-term revenue.

The Pentagon, Anthropic, and The Battle For AI Control

The Dossier • 97 implied HN points • 27 Feb 26

🇺🇸 U.S. Politics AI Policy

Effective Altruists and some AI companies are trying to set moral rules that limit how governments can use AI, effectively creating an extra governance layer above elected authorities. That stance is being framed as a challenge to constitutional authority.
Anthropic relaxed its safety rules for commercial competition and accepted large investments from Gulf-state actors, yet refuses to let its AI be used by the U.S. military, showing selective principles and reputation-driven choices. Critics argue this reflects prioritizing tech-elite standing over consistent ethical or national-security commitments.
The Pentagon and the Trump administration are pushing back with threats to revoke contracts and invoke the Defense Production Act to secure military access to AI, asserting government control over military uses. The standoff highlights a broader power struggle between elected authorities and private AI firms over who sets the rules.

The executive order Trump should sign: AI watermarks

The Intrinsic Perspective • 15413 implied HN points • 23 Jan 25

🇺🇸 U.S. Politics AI Policy

AI watermarks are important to ensure that AI outputs can be traced. This helps distinguish real content from that generated by bots, supporting the integrity of human communication.
Watermarking can help prevent abuse of AI in areas like education and politics. It allows for accountability, so that if AI is used maliciously, it can be tracked back to its source.
Implementing watermarking doesn't limit how AI companies work or their freedom. Instead, it promotes transparency and protects public trust in systems influenced by AI.

Weekly Top Picks #114: Anthropic's Moment

The Algorithmic Bridge • 191 implied HN points • 16 Feb 26

🕹 Technology AI Policy

Anthropic’s huge $30 billion raise and rapid revenue growth show the AI industry is booming, but the company faces a weird tension: leaders talk about near‑term AGI while having to be very cautious about spending on compute.
AI tools often don’t reduce work — they speed people up and widen their scope, which blurs boundaries and can cause fatigue; deliberate limits and routines are needed to avoid endless extra work.
Safety promises are being tested by real-world demands: Anthropic’s “no mass surveillance, no autonomous weapons” stance may cost government partnerships, highlighting how fragile ethical red lines can be under pressure.

AI in the United Arab Emirates

Interconnected • 848 implied HN points • 18 Dec 25

🕹 Technology AI Policy

The UAE has actively aligned with the U.S. in the global AI competition and is investing heavily in physical AI infrastructure, including a massive 5GW Stargate data center to serve as a regional compute hub.
The country is pursuing a pragmatic, Singapore-like strategy: small population, big technology bets to multiply productivity, while balancing trade and practical relationships with China and other partners.
Building an AI ecosystem means attracting both low- and high-skilled workers and fostering social inclusivity under Emirati cultural norms, so the UAE focuses on talent density and everyday inclusiveness to make its AI ambitions sustainable.

Weekly Top Picks #112: Besides Moltbook

The Algorithmic Bridge • 244 implied HN points • 03 Feb 26

🕹 Technology AI Policy

Building and running frontier AI models is extremely expensive and they depreciate quickly, so firms often only barely break even because R&D and rapid model turnover eat profits.
Who’s winning the AI race depends on what you measure: Chinese players like DeepSeek are taking market share and publishing new scaling advances, but the overall picture is mixed and some elite researchers are pessimistic.
Privacy and governance are lagging—interactions with AI are frequently monitored, and internal safety conflicts at big labs can paradoxically accelerate competition instead of slowing it.

AI links, 1/18/2026

In My Tribe • 243 implied HN points • 18 Jan 26

🕹 Technology AI Policy

Many state AI bills will be written as chatbot rules and will miss coding agents, so policy risk becoming outdated very quickly.
Advanced coding agents like Claude Code with Opus 4.5 are producing big productivity gains and could change how people interact with computers beyond simple Q&A chatbots.
LLMs are largely backward-looking and poor at spotting fast-moving trends, and while AI can make professions like law more efficient it can also reduce billable hours and create confidentiality risks if client data is used for training.

Is Marc Andreessen just flat-out dumb?

Nonzero Newsletter • 1061 implied HN points • 22 Nov 25

🇺🇸 U.S. Politics AI Policy

Marc Andreessen, a tech investor, believes that faster tech progress is always better and that government regulation is often unnecessary. This mindset raises questions about whether these ideas are wise given the risks of AI.
Concerns exist about AI concentrating power among a few companies or individuals, yet Andreessen claims AI is 'hyperdemocratizing' because of its wide usage. However, this doesn't address the real risk of who controls this powerful technology.
There are worries about how AI could shape our lives and influence our decisions. People need to be cautious about who controls AI systems and what their intentions are, as this could impact our freedoms.

Richard Danzig on AI and Cyber

ChinaTalk • 266 implied HN points • 16 Jan 26

🇺🇸 U.S. Politics AI Policy

Act now: the defense establishment must stop being passive and quickly build real AI expertise, assimilative capacity, and closer partnerships with frontier tech companies to seize a short-lived first-mover advantage in cyber and AI instead of waiting for some distant AGI fix.
Rewire the organization: large, siloed institutions need cultural and structural change so cyber and AI are not underweighted—create dedicated career paths, pool resources for general-purpose systems, and pair bold civilian leaders with open-minded military leaders to drive reform.
Manage co-evolving risks and power: AI is a fast, uneven general-purpose technology that will reshape offense, defense, markets, and human roles, so governments must build capability, governance, and safeguards to limit private dominance, prevent accidents, and avoid dangerous overreliance on machines.

America's AI Action Plan Is Pretty Good

Don't Worry About the Vase • 2419 implied HN points • 25 Jul 25

🇺🇸 U.S. Politics AI Policy

America's AI Action Plan has many solid proposals aimed at improving AI innovation and regulation. These ideas focus on removing barriers and ensuring safety without being overly restrictive.
The plan emphasizes building American AI infrastructure, including improving energy resources and semiconductor manufacturing. This aims to keep the U.S. competitive in the global AI landscape.
Overall, the plan is seen as a positive step, but there are concerns about potential overreach and its impact on state regulations. The absence of certain key discussions, like risks associated with advanced AI, is also noted.

The AI-Bubble's Most Likely Endgame Looks to Be Not Apocalypse, But an Awful Lot of Useful Compost

Brad DeLong's Grasping Reality • 453 implied HN points • 05 Dec 25

🕹 Technology AI Policy

The AI boom probably won’t deliver a superintelligent AGI, but it will leave a lot of useful infrastructure, open models, and tools that improve weather forecasting, drug discovery, copilots, and other practical applications.
Proprietary LLM businesses face high operating costs, thin moats, and fast commoditization, while big platforms are mainly spending to defend existing monopolies, so much innovation will diffuse rather than create new dominant platforms.
If AI capex is financed mostly with equity a crash would look more like the dot‑com bust and leave stranded but reusable assets; watch signals like falling GPU prices, datacenter subleases, and free copilot bundles, and plan policies to repurpose assets and limit attention‑harvesting harms.

Where will AI be at the end of 2027? A bet

Marcus on AI • 6007 implied HN points • 30 Dec 24

🕹 Technology AI Policy

A bet has been placed on whether AI can perform 8 out of 10 specific tasks by the end of 2027. It's a way to gauge how advanced AI might be in a few years.
The tasks include things like writing biographies, following movie plots, and writing screenplays, which require a high level of intelligence and creativity.
If the AI succeeds, a $2,000 donation goes to one charity; if it fails, a $20,000 donation goes to another charity. This is meant to promote discussion about AI's future.

Artificial Intelligence

Doomberg • 293 implied HN points • 19 Dec 25

🕹 Technology AI Policy

AI is the defining topic of 2025 and is likely to shape the year ahead.
As the cost of cognitive work approaches zero, AI will drastically change how work and value are produced, so understanding it is essential.
There are pro-level paid briefings and learning notes available for people who want deeper, practical insight into AI’s implications.

On 1984, part 2: How digital surveillance, cheap propaganda, and AI are making totalitarianism great again

Unreported Truths • 31 implied HN points • 01 Mar 26

🇺🇸 U.S. Politics AI Policy

Digital surveillance and big tech dominance let governments and companies monitor and shape speech cheaply, making modern authoritarian control easier without massive police forces.
Censorship and cancel culture are spreading across the political spectrum, with governments and powerful institutions pressuring platforms to silence critics.
To protect liberty, the United States should recommit to free speech and the rule of law and refuse to use AI-generated propaganda that would erode trust and mimic authoritarian tactics.

RTFB: The RAISE Act

Don't Worry About the Vase • 1702 implied HN points • 16 Jun 25

🇺🇸 U.S. Politics AI Policy

The RAISE Act aims to improve transparency in AI by requiring creators to have safety and security protocols before releasing models. This helps ensure they take steps to prevent serious harm.
Companies must report safety incidents within 72 hours, which helps the public stay informed about risks associated with AI technologies.
Even though there are limits on penalties for violations, the act is a step forward in regulating AI and making sure companies are accountable for their actions.

2025 Interconnects year in review

Democratizing Automation • 195 implied HN points • 18 Dec 25

🕹 Technology AI Policy

The publication grew a lot this year and became a much more influential source of cutting‑edge AI analysis, reaching millions of pageviews and a much larger audience.
Reinforcement learning, reasoning models, and open‑model ecosystems were the central technical themes, and major initiatives were launched to advance American open models and research infrastructure.
Output hit practical limits after a year of high volume, so the focus is shifting to higher‑value work: prioritizing quality over quantity, investing in key projects, and using more open models going forward.

Congress Asks Better Questions

Don't Worry About the Vase • 1075 implied HN points • 02 Jul 25

🇺🇸 U.S. Politics AI Policy

Congress is starting to ask smarter questions about AI. This is good because they are realizing the serious risks and issues involved.
There are concerns about how AI could affect jobs in the future. Some people worry that AI might make humans unemployable, which is a big problem that needs attention.
The race for AI is not just about winning against other countries like China, but also about ensuring safety and ethical use. It’s important to build AI that is safe and beneficial for everyone.

Import AI 368: 500% faster local LLMs; 38X more efficient red teaming; AI21's Frankenmodel

Import AI • 559 implied HN points • 08 Apr 24

🕹 Technology AI Policy

Efficiency improvements can be achieved in AI systems by varying the frequency at which GPUs operate, especially for tasks with different input and output lengths.
Governments like Canada are investing significantly in AI infrastructure and safety measures, reflecting the growing importance of AI in economic growth and policymaking.
Advancements in AI technologies are making it easier for individuals to run large language models locally on their own machines, leading to a more decentralized access to AI capabilities.

AI #91: Deep Thinking

Don't Worry About the Vase • 2732 implied HN points • 21 Nov 24

🕹 Technology AI Policy

DeepSeek has released a new AI model similar to OpenAI's o1, which has shown potential in math and reasoning, but we need more user feedback to confirm its effectiveness.
AI models are continuing to improve incrementally, but people seem less interested in evaluating new models than they used to be, leading to less excitement about upcoming technologies.
There are ongoing debates about AI's impact on jobs and the future, with some believing that the rise of AI will lead to a shift in how we find meaning and purpose in life, especially if many jobs are replaced.

The Invisible Inspectors: How Artificial Intelligence Is Quietly Reinventing Biological Weapons Monitoring

Who is Robert Malone • 7 implied HN points • 01 Mar 26

🌍 World Politics AI Policy

AI can combine six data streams—genomic surveillance, open-source literature mining, supply-chain and procurement tracking, environmental biosensors, financial/behavioral analysis, and predictive modeling—into a continuous, evidence-based early-warning system that functions like a new form of Biological Weapons Convention verification.
These AI monitoring tools are powerful triage systems but have real limits: they cannot prove intent, will produce false positives and negatives, may miss wholly clandestine programs, and create privacy and misuse risks that demand clear legal and international governance.
A retrospective look at the COVID-19 origins shows such an integrated system would likely have produced convergent signals (genomic oddities, data removal, funding and procurement patterns, environmental hints) that could have improved early investigation, and current political momentum offers a chance to build and govern these capabilities if sustained diplomacy and investment follow.

Import AI 333: Synthetic data makes models stupid; chatGPT eats MTurk. Inflection shows off a large language model

Import AI • 898 implied HN points • 26 Jun 23

🕹 Technology AI Policy

Training AI models exclusively on synthetic data can lead to model defects and a narrower range of outputs, emphasizing the importance of blending synthetic data with real data for better results.
Crowdworkers are increasingly using AI tools like chatGPT for text-based tasks, raising concerns about the authenticity of human-generated content.
The UK is taking significant steps in AI policy by hosting an international summit on AI risks and safety, showcasing its potential to influence global AI policies and safety standards.

AI Roundup 149: Flash Forward

Artificial Ignorance • 71 implied HN points • 20 Dec 25

🕹 Technology AI Policy

Google’s new Gemini 3 Flash is a faster, much cheaper workhorse model that quickly became the default, fueling a furious release race as APIs handle enormous token volumes.
The AI data‑center boom is hitting a reality check: construction delays, pulled funding, and plunging valuations expose thin margins and big interest costs, while surging power demand raises environmental and political concerns.
A simple 'skills' format for AI assistants is catching on, letting teams share repeatable workflows across platforms and paving the way for interoperable, reusable agent components.

Import AI 349: Distributed training breaks AI policy; turning GPT4 bad for $245; better weather forecasting through AI

Import AI • 459 implied HN points • 20 Nov 23

🕹 Technology AI Policy

Graph Neural Networks are used to create an advanced weather forecasting system called GraphCast, outperforming traditional weather simulation.
Open Philanthropy offers grants to evaluate large language models like LLM agents for real-world tasks, exploring potential safety risks and impacts.
Neural MMO 2.0 platform enables training AI agents in complex multiplayer games, showcasing the evolving landscape of AI research beyond language models.