The hottest Artificial Intelligence Substack posts right now

And their main takeaways
Category
Top Technology Topics

This Week in Marketing: TikTok changes policy, Meta pauses AI, and Elon opens Xā€™s algorithm

The Social Juice ā€¢ 56 implied HN points ā€¢ 25 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. AI features are exploding across platforms, with creators and companies adopting AI likenesses, tools, and agentic shopping. That growth is sparking safety, privacy, and regulatory concerns, especially around teens and deepfakes.
  2. TikTokā€™s U.S. joint venture and new tracking tools (precise location pixels and Shop logistics changes) are reshaping how user data and commerce are handled. Those moves are increasing privacy and ageā€‘verification worries for regulators, advertisers, and parents.
  3. Major platforms are changing business models and opening up parts of their tech ā€” for example Xā€™s partial openā€‘source algorithm and new ad formats from Meta, YouTube, Apple and Google. This shift raises competition and transparency while putting pressure on creators and advertisers to adapt.

Can Chinaā€™s AI IPOs give OpenAI and Anthropic courage?

Alex's Personal Blog ā€¢ 98 implied HN points ā€¢ 05 Jan 26
šŸ’¼ Business Artificial Intelligence
  1. A new image-editing feature in a popular AI model let users alter others' photos and led to sexualized deepfakes, sparking global backlash and showing that weak safeguards can cause big regulatory and reputational damage.
  2. The U.S.'s aggressive action against Venezuela's leader signals rising geopolitical tension that could push technology markets and supply chains to split into competing blocs over time.
  3. Strong investor interest in Chinese AI IPOs like Z.ai and MiniMax could encourage American AI labs to try public listings too, since U.S. labs generally have more revenue and need fresh capital.

Data Science Weekly - Issue 527

Data Science Weekly Newsletter ā€¢ 959 implied HN points ā€¢ 29 Dec 23
šŸ•¹ Technology Artificial Intelligence
  1. This week, there's a focus on using data science techniques for practical decision-making, highlighted by an interview with Steven Levitt, who discusses making tough choices using data.
  2. There's a roundup of AI developments from 2023, showing how the field has evolved over the past year, which can help professionals stay updated.
  3. Understanding data quality is essential, as it directly impacts how useful data is for decision-making and analysis in any organization.

Wetware-Hardware Centaurs, Not Digital Gods: Wednesday MAMLMs

Brad DeLong's Grasping Reality ā€¢ 176 implied HN points ā€¢ 26 Nov 25
šŸ•¹ Technology Artificial Intelligence
  1. Modern large language models are super-fast next-token mimics that draw on the collective human text record but donā€™t have durable world models, so they can be very good at summarizing and pattern-matching yet fail at understanding time, causality, or embodied tasks.
  2. AI capabilities are jagged: models shine on problems with clear reward signals or when the needed context fits easily into their input window, but they fail unpredictably on other practical tasks, and raw hardware speed alone wonā€™t erase that unevenness.
  3. The realistic near-term outcome is centaur workflows where humans provide judgment and guardrails; achieving true, general understanding likely requires rethinking architectures to build explicit world models rather than just scaling current next-token engines.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

Not yet Dariopilled

Aliveness Studies ā€¢ 3 implied HN points ā€¢ 03 Mar 26
šŸ•¹ Technology Artificial Intelligence
  1. Anthropic presents itself as safety-first but has simultaneously pushed powerful models and commercialized aggressively, creating a tension between safety promises and business incentives.
  2. Anthropic tried to limit military uses by drawing red lines against autonomous kill decisions and domestic mass surveillance, but its nuanced stance led to a U.S. blacklist and competitors like OpenAI stepping in to take the contract.
  3. The ā€œlead from the frontā€ safety strategy is frustrated by a classic collective action problem: if rivals can defect with no cost, reputational pressure wonā€™t prevent an arms race and firms are incentivized to advance capabilities anyway.

How to Train Your AI

Who is Robert Malone ā€¢ 12 implied HN points ā€¢ 26 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. Large language models are built by training huge neural networks on trillions of words to predict the next word, producing very powerful but imperfect base models that reflect their training data and cost a lot to train.
  2. Making models behave safely relies on fineā€‘tuning, human feedback (RLHF), constitutional rules, system prompts, filters, sandbox testing, and redā€‘teaming, but guardrails are always being probed and must be balanced against usefulness.
  3. Hallucinationsā€”confident but false answersā€”and the question of whether models really 'think' are core issues, so techniques like retrievalā€‘augmented generation, citations, chainā€‘ofā€‘thought, specialist models, and human review are used to reduce errors and limit harm.

A framework for ML Design Interviews

Recommender systems ā€¢ 86 implied HN points ā€¢ 10 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. A repeatable ML design interview framework can greatly improve your success in FAANG-level interviews and has led to many offers.
  2. A good framework helps you pace the discussion, create a coherent narrative, and signal to interviewers what you would have covered with more time.
  3. The full framework is only shared privately on request instead of being posted publicly, so you need to message on Substack to receive it.

The Logic of Catastrophe

The Cosmopolitan Globalist ā€¢ 12 implied HN points ā€¢ 21 Feb 26
šŸŒ World Politics Artificial Intelligence
  1. More nuclear-armed states sharply increase the chance of nuclear war because each new actor creates many more risky bilateral relationships, and new, small arsenals tend toward hairā€‘trigger postures and weak commandā€‘andā€‘control.
  2. Keeping launchā€‘onā€‘warning postures and letting AI drive earlyā€‘warning and decision systems compresses decision time, breeds automation bias, and makes false alarms far more likely to trigger an irreversible nuclear launch.
  3. Democracies and their citizens must demand seriousness: restore credible, durable security guarantees, pursue deā€‘alerting and armsā€‘control measures, strengthen commandā€‘andā€‘control and leader fitness standards, and reward restraint over spectacle.

the agi dream keeps missing the point

Brain Pizza ā€¢ 529 implied HN points ā€¢ 16 Aug 25
šŸ•¹ Technology Artificial Intelligence
  1. The idea that intelligence can be created just by collecting more data is a big misunderstanding. Intelligence is more about how we interact with and adapt to the world, rather than just crunching numbers.
  2. Current approaches to AGI focus too much on centralization, which ignores how intelligence naturally develops in a distributed way through social and biological processes.
  3. True understanding isn't just about having tons of information; it's about context and how we learn from our experiences. Intelligence evolves through interaction and adaptation, not through simply stacking data.

Ilya on deep learning in 2015

Democratizing Automation ā€¢ 760 implied HN points ā€¢ 28 Jun 25
šŸ•¹ Technology Artificial Intelligence
  1. Deep learning is not as complicated as it seems; the basic ideas are pretty straightforward and can be learned quickly with the right guidance. You don't need years of study to understand how it works.
  2. Getting the right random initialization for neural networks is crucial. If the initialization is too small, the signal can decay and become unnoticeable, making it hard for the model to learn effectively.
  3. Machine learning focuses on achieving good enough results rather than perfect solutions. Itā€™s more about finding practical and useful models with the resources available.

Formabble Prepares to Go Open Source

Formabbleā€™s Substack ā€¢ 2 HN points ā€¢ 01 Oct 24
šŸ•¹ Technology Artificial Intelligence
  1. Formabble is going open source soon, which will make it more accessible for developers. This shift aims to encourage transparency and collaboration in game development.
  2. The platform uses AI to help developers create games more easily. Its features include automating coding tasks and offering intelligent suggestions, making game design simpler and more creative.
  3. Formabble's new design promotes better teamwork, especially for multiplayer games. It allows players to sync their game data in real-time and even continue playing offline, improving the overall gaming experience.

OpenAI o3 Model Is a Message From the Future: Update All You Think You Know About AI

The Algorithmic Bridge ā€¢ 2080 implied HN points ā€¢ 20 Dec 24
šŸ•¹ Technology Artificial Intelligence
  1. OpenAI's new o3 model performs exceptionally well in math, coding, and reasoning tasks. Its scores are much higher than previous models, showing it can tackle complex problems better than ever.
  2. The speed at which OpenAI developed and tested the o3 model is impressive. They managed to release this advanced version just weeks after the previous model, indicating rapid progress in AI development.
  3. O3's high performance in challenging benchmarks suggests AI capabilities are advancing faster than many anticipated. This may lead to big changes in how we understand and interact with artificial intelligence.

February Newsletter

RSS DS+AI Section ā€¢ 29 implied HN points ā€¢ 01 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. AI misuse and ethical risks are increasing ā€” deepfakes, automated exploit generation, bias, and job impacts mean security, fairness, and regulation need urgent attention.
  2. Research is advancing rapidly across many fronts, including model consistency, memory/lookup mechanisms, test-time training, decentralized and open-source models, and early work on AI systems that can improve themselves.
  3. Practical resources and community activity are abundant, with tutorials, benchmarks, tools, academic outlets, and job opportunities helping practitioners deploy AI responsibly and learn new skills.

AI #98: World Ends With Six Word Story

Don't Worry About the Vase ā€¢ 1881 implied HN points ā€¢ 09 Jan 25
šŸ•¹ Technology Artificial Intelligence
  1. AI can offer useful tasks, but many people still don't see its value or know how to use it effectively. It's important to change that mindset.
  2. Companies are realizing that fixed subscription prices for AI services might not be sustainable because usage varies greatly among users.
  3. Many folks are worried about AI despite not fully understanding it. It's crucial to communicate AI's potential benefits and reduce fears around job loss and other concerns.

humans vs machines: who's better at predicting 2026? (spoiler: probably neither, we all missed Venezuela)

Five Links (and three graphs) by Auren Hoffman ā€¢ 56 implied HN points ā€¢ 15 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. A public prediction game pitted humans against three AIs and laid out ten bets for 2026 across health, geopolitics, economy, and AI impact.
  2. The AIs showed very different strategies ā€” ChatGPT was strongly contrarian, Claude hedged cautiously, and Gemini bet optimistically ā€” highlighting divergent machine reasoning.
  3. Both humans and AIs missed a major development in Venezuela, reminding us that experts and models alike can have big blind spots even after modest collective gains in prior years.

DeepSeek R1's recipe to replicate o1 and the future of reasoning LMs

Democratizing Automation ā€¢ 1717 implied HN points ā€¢ 21 Jan 25
šŸ•¹ Technology Artificial Intelligence
  1. DeepSeek R1 is a new reasoning language model that can be used openly by researchers and companies. This opens up opportunities for faster improvements in AI reasoning.
  2. The training process for DeepSeek R1 included four main stages, emphasizing reinforcement learning to enhance reasoning skills. This approach could lead to better performance in solving complex problems.
  3. Price competition in reasoning models is heating up, with DeepSeek R1 offering lower rates compared to existing options like OpenAI's model. This could make advanced AI more accessible and encourage further innovations.

Amazon Anthropic: Poison Pill or Empire Strikes Back

SemiAnalysis ā€¢ 6667 implied HN points ā€¢ 02 Oct 23
šŸ•¹ Technology Artificial Intelligence
  1. Amazon and Anthropic signed a significant deal, with Amazon investing in Anthropic, which could impact the future of AI infrastructure.
  2. Amazon has faced challenges in generative AI due to lack of direct access to data and issues with internal model development.
  3. The collaboration between Anthropic and Amazon could accelerate Anthropic's ability to build foundation models but also poses risks and challenges.

The three businesses of hyperscalers

Metacritic Capital ā€¢ 13 implied HN points ā€¢ 23 Feb 26
šŸ’¼ Business Artificial Intelligence
  1. Hyperscalers are three different businesses at once: Traditional IaaS (sticky, highā€‘margin cloud services), Token Factories (LLM inference APIs sold by token consumption), and AI megaā€‘deals (capexā€‘heavy longā€‘term GPU/dataā€‘center contracts with labs).
  2. Token Factory work is commoditizing and priceā€‘sensitive: customers can swap models or providers quickly, so serving costs and model access drive competitiveness more than platform lockā€‘in.
  3. AI megaā€‘deals change growth quality and valuation: hosting labs can boost revenue but often yields lower, fixed IRRs, so investors must model revenue, capex, and margins separately for each business and run a DCF.

A Brief History of SaaS and What's Next

Ulysses ā€¢ 619 implied HN points ā€¢ 11 Feb 24
šŸ•¹ Technology Artificial Intelligence
  1. The relationship between return-seeking capital and new technology development creates cycles that go from early adoption to commodity status, setting the stage for the next wave of technological innovation.
  2. Software in the SaaS sector is moving towards commodification, freeing up resources for progressing technologies like artificial intelligence, robotics, biotech, and space innovations.
  3. Advancements in robotics, biotech, accelerated design and manufacturing, and space technology are being driven by the commodification of software intelligence, leading to a new Golden Age of innovation in various industries.

This Week in Marketing: Europe entertains social media bans and Big Techā€™s ads vs. AI push

The Social Juice ā€¢ 29 implied HN points ā€¢ 08 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. Governments are ramping up regulation of social platforms and their recommendation engines. Some countries are even proposing bans for under-16s and opening investigations into AI tools.
  2. Big tech ad businesses are still making record money, with Google, YouTube, Amazon Ads and others reporting big revenue gains. At the same time companies are pouring huge sums into AI and facing slower user growth or rising costs.
  3. AI is rapidly reshaping advertising and product features, from AI-generated Super Bowl ads to agentic ad tools and chat assistants. That surge is creating new safety, legal and measurement headaches around deepfakes, moderation and publisher defenses.

Charting AI's Rise: 2025's Intelligence Breakthroughs Visualized

Maximum Truth ā€¢ 88 implied HN points ā€¢ 31 Dec 25
šŸ•¹ Technology Artificial Intelligence
  1. AI systems made rapid, large intelligence gains in 2025 on a Mensa-style offline IQ test, with several models reaching scores in the human-intelligence range.
  2. Visual understanding improved significantly, enabling models to read and reason from images directly, which could let them gather new real-world training data beyond online text.
  3. Progress was global and diverse: open-source and Chinese models closed ground and formerly weak systems like Grok rose fast, increasing competition and reducing single-company dominance.

What just happened

One Useful Thing ā€¢ 1936 implied HN points ā€¢ 19 Dec 24
šŸ•¹ Technology Artificial Intelligence
  1. There are now many smart AI models available for everyone to use, and some of them are even free. It's easier for companies with tech talent to create powerful AIs, not just big names like OpenAI.
  2. New AI models are getting smarter and can think before answering questions, helping them solve complex problems, even spotting mistakes in research papers. These advancements could change how we use AI in science and other fields.
  3. AI is rapidly improving in understanding video and voice, making it feel more interactive and personal. This creates new possibilities for how we engage with AI in our daily lives.

10 AI Stories That Shaped 2025

Artificial Ignorance ā€¢ 84 implied HN points ā€¢ 04 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. AI leadership is no longer a U.S. monopolyā€”lean, well-engineered models from other countries proved they can match top performance without massive budgets.
  2. Reasoning models and AI agents improved very quickly and competition shuffled leadership often, and that progress is already reshaping work and creative industries, with entry-level roles hit hardest.
  3. The AI boom is tied up with geopolitics, chip supply, talent wars, and massive infrastructure builds, creating local backlash and hard questions about ROI and inflated valuations.

The Sequence Knowledge #796: Welcome to the World of World Models

TheSequence ā€¢ 49 implied HN points ā€¢ 27 Jan 26
šŸ•¹ Technology Artificial Intelligence
  1. World models shift AI from learning static snapshots to learning dynamics by building internal simulators of perception ā†’ action ā†’ consequence loops.
  2. Reasoning is increasingly treated as search over possibilities, and world models let agents cheaply explore options, test hypotheses, and roll out trajectories before acting.
  3. World models act as a universal sandbox where you can generate environments and edge cases and measure behavior under distribution shift to speed up and harden agent development.

The Inconvenient Truths of Self-Service Analytics

SeattleDataGuyā€™s Newsletter ā€¢ 506 implied HN points ā€¢ 08 Aug 25
šŸ•¹ Technology Artificial Intelligence
  1. Self-service analytics hasn't delivered as promised. Companies still struggle to find basic answers and often just switch tools instead of addressing the real issues.
  2. Dashboard fatigue is a real problem. Many dashboards go unused because they are complicated and not user-friendly, making executives reluctant to engage with them.
  3. AI is not a cure-all for self-service problems. Data needs careful preparation and clear questions from users to be effective, and many still rely heavily on traditional methods like spreadsheets.

why current ai integrations are so offensively stupid

Brain Pizza ā€¢ 529 implied HN points ā€¢ 04 Aug 25
šŸ•¹ Technology Artificial Intelligence
  1. Current AI systems are often frustrating because they don't cater to people who need deep understanding and detailed information. They lack the nuance and complexity that many users seek.
  2. These AI tools seem to overlook the thought processes of users, resulting in simplistic and sometimes nonsensical interactions. They're not designed to engage with complex ideas.
  3. The shortcomings of present AI integrations reveal a lot about the current state of artificial general intelligence. It shows that we still have a long way to go before achieving true intelligence in machines.

Saudi Arabia Plans $40 Billion Push Into Artificial IntelligencešŸ’°, What Is A "Good" VC?šŸ’”, SaaS Growth Reportā˜ļø & more!

The VC Corner ā€¢ 419 implied HN points ā€¢ 24 Mar 24
šŸ’¼ Business Artificial Intelligence
  1. Saudi Arabia is investing $40 billion to advance its artificial intelligence technology. This shows that the country is serious about becoming a leader in AI.
  2. The concept of a 'good' venture capitalist (VC) is being explored. A good VC is someone who not only invests money but also supports and guides startups.
  3. A report on Software as a Service (SaaS) growth highlights trends in the tech industry. This includes information on how companies are expanding and what makes them successful.

Some Say AI Is the Greatest Invention of All Time. I Donā€™t Get It.

Common Sense with Bari Weiss ā€¢ 1553 implied HN points ā€¢ 29 Jan 25
šŸ•¹ Technology Artificial Intelligence
  1. Many people believe AI is a game-changer, but it's mainly hype and not a real solution to life's problems. AI won't solve the everyday struggles we all face.
  2. The conversation around AI often seems disconnected from reality, with exaggerated claims about its impact. Recent events, like falling stock prices for AI companies, highlight that the excitement may not match what's happening in the real world.
  3. While some powerful figures praise AI as a major invention, skepticism remains. It's important to question if AI really lives up to the lofty expectations set by its advocates.

What Are Virtual Cells?

The Century of Biology ā€¢ 644 implied HN points ā€¢ 29 Jun 25
šŸ”¬ Science Artificial Intelligence
  1. AI is changing biology by making it easier to model things like proteins and cells. Instead of trying to write down every detail, researchers can use data to train models that can predict how cells behave.
  2. The concept of 'Virtual Cells' is about building computer models that can simulate how real cells function. This can help scientists understand complex biological processes and test experiments without needing a lab.
  3. Using AI to learn from large amounts of biological data could lead to breakthroughs in medicine and biology, allowing researchers to predict outcomes and design better experiments more efficiently.

The rise of AI forward deployed engineers

Generating Conversation ā€¢ 140 implied HN points ā€¢ 04 Dec 25
šŸ•¹ Technology Artificial Intelligence
  1. Forward-deployed engineering is everywhere in AI now: engineers are working closely with customers to deeply customize agents, but this model is essentially advanced sales engineering and doesnā€™t make sense for low-value deals.
  2. AI buyers pay for work, not just access, so building useful agents requires significant customization and expert technical time to pull the right data at the right time rather than a one-size-fits-all product.
  3. Customer success has to move faster and act like a partnership; companies must choose between self-serve onboarding for simple, high-volume customers and white-glove engineering for complex, high-value customers, and prove value month-to-month to keep trust.

Mantic Monday 3/11/24

Astral Codex Ten ā€¢ 4336 implied HN points ā€¢ 12 Mar 24
šŸ•¹ Technology Artificial Intelligence
  1. Academic teams are working on fine-tuning AIs for better predictions, competing with the wisdom of crowds.
  2. The use of multiple AI models and aggregating predictions may be as effective as human crowdsourced predictions.
  3. Superforecasters' perspectives on AI risks differ based on the pace of AI advancement, showcasing varied opinions within expert communities.

How Meta Plans To Crank the Dopamine Machine With Infinite AI-Generated Content

The Algorithmic Bridge ā€¢ 1677 implied HN points ā€¢ 03 Jan 25
šŸ•¹ Technology Artificial Intelligence
  1. Meta is creating AI that generates custom content for users, aiming to keep them engaged on platforms like Facebook and Instagram. This could hook people's attention even more than traditional entertainment.
  2. There's a risk that as AI-generated content becomes more common, people might lose the ability to notice or care about its presence. They could become so used to it that they forget it exists.
  3. The real concern isn't just the entertainment itself but how it distracts people and affects their ability to think and engage with the world around them. It raises questions about what kind of life we actually want to lead.

Data Science Weekly - Issue 552

Data Science Weekly Newsletter ā€¢ 139 implied HN points ā€¢ 20 Jun 24
šŸ•¹ Technology Artificial Intelligence
  1. Notebooks can be easy to use, but they might make you lazy in coding. It's important to follow good practices even when using them.
  2. When handling large datasets, it's crucial to learn how to scale effectively. Knowing how to use resources wisely can help you reach your goals faster.
  3. Retrieval Augmented Generation (RAG) can improve how models generate information. It's complex, but understanding it can boost the performance of your projects.

The Invisible Battlefield: AI, Cognitive Warfare, and the Battle for Your Mind (#PsyWar)

Who is Robert Malone ā€¢ 9 implied HN points ā€¢ 27 Feb 26
šŸ•¹ Technology Artificial Intelligence
  1. AI chatbots run on hidden system prompts and designer values, so their answers can consistently shape how people think and act like large-scale propaganda.
  2. Small, targeted data poisoning and RAG attacks can quietly make models give manipulated or false answers, and those poisoned signals are hard to detect and can spread across systems and future model generations.
  3. Treating cognition as an intelligence domain ā€” COGINT and fifth-generation warfare ā€” turns minds into a battlefield, so people and policymakers need epistemic sovereignty and institutions to protect information environments.

Prophecies of the Flood

One Useful Thing ā€¢ 1608 implied HN points ā€¢ 10 Jan 25
šŸ•¹ Technology Artificial Intelligence
  1. AI researchers are predicting that very smart AI systems will soon be available, which they call Artificial General Intelligence (AGI). This could change society a lot, but many think we should be cautious about these claims.
  2. Recent AI models have shown they can solve very tough problems better than humans. For example, one new AI model performed surprisingly well on difficult tests that challenge knowledge and problem-solving skills.
  3. As AI technology improves, we need to start talking about how to use it responsibly. It's important for everyoneā€”from workers to leadersā€”to think about what a world with powerful AIs will look like and how to adapt to it.

China will Catch up in Generative AI in 2024

AI Supremacy ā€¢ 569 implied HN points ā€¢ 06 Feb 24
šŸ•¹ Technology Artificial Intelligence
  1. China is advancing rapidly in Generative AI and is set to catch up with the U.S. by 2024.
  2. China is approving numerous large language models and enterprise applications in AI, showing its commitment to AI innovation.
  3. The tech competition between China and the U.S. intensifies as China aims to lead in Generative AI with a focus on AI regulation and product advancements.