The hottest AI Substack posts right now

And their main takeaways
Category
Top Technology Topics

The Weekly Kaitchup #64

The Kaitchup – AI on a Budget • 59 implied HN points • 25 Oct 24
đź•ą Technology AI
  1. Qwen2.5 models have been improved and now come in a 4-bit version, making them efficient for different hardware. They perform better than previous models on many tasks.
  2. Google's SynthID tool can add invisible watermarks to AI-generated text, helping to identify it without changing the text's quality. This could become a standard practice to distinguish AI text from human writing.
  3. Cohere has launched Aya Expanse, new multilingual models that outperform many existing models. They took two years to develop, involving thousands of researchers, enhancing language support and performance.

Now I Really Won That AI Bet

Astral Codex Ten • 30146 implied HN points • 08 Jul 25
đź•ą Technology AI
  1. In 2022, a bet was made on whether AI could create complex images by 2025. The challenge was to generate images that matched detailed prompts.
  2. Over the years, various AI models were tested, and the results showed both progress and limitations. Improvements were made, but some details were still missed.
  3. By June 2025, an updated AI model finally met all the conditions of the bet, showing that AI can achieve a high level of image generation based on specific instructions.

Program or Be Programmed - Launch Party

Rushkoff • 199 implied HN points • 17 Oct 24
đź•ą Technology AI
  1. There is a book launch party happening in NYC on November 3, celebrating the updated edition of 'Program or Be Programmed.'
  2. The event includes a conversation about the impact of psychedelics and digital society's future.
  3. Attendance is free for a limited number of people who RSVP, and it will also be live-streamed for those who can't attend in person.

⏸️💨 The pause before the gale

Faster, Please! • 1462 implied HN points • 06 Feb 26
đź•ą Technology AI
  1. AI is currently creeping into many jobs and industries unevenly, but its technical capabilities are improving fast and could trigger a sudden, much bigger shift down the road.
  2. The short-term picture is mixed: some firms will see big productivity gains while many workers and incumbent businesses face disruption, and public anxiety can amplify market volatility.
  3. If companies invest more in data, systems integration, and reorganizing work, AI could move beyond automating tasks to raise overall productivity and unlock large gains in growth, wages, health, and education.

Advancements In Self-Driving Cars

Don't Worry About the Vase • 3136 implied HN points • 07 Jan 26
đź•ą Technology AI
  1. Waymo is rapidly expanding driverless service across many cities and freeways, but growth depends on getting more vehicles and clearing state and local regulatory hurdles.
  2. Autonomous cars are already much safer than human drivers and act cautiously in events like power outages, yet those incidents show the need for better protocols and sensible rule changes (for example on speed limits).
  3. Widespread self-driving will reshape daily life—giving huge benefits to cyclists, the elderly, and deliveries while disrupting driving jobs—so policy choices must manage those social and economic impacts.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

33 Things I Heard At Foo Camp 2026

The Ruffian • 436 implied HN points • 28 Feb 26
đź•ą Technology AI
  1. Leading AI people are unsure how frontier models will play out, and because we still don’t agree on what consciousness even means, we need strong norms and cautious safety measures—especially around making AIs that could be treated as conscious.
  2. Modern reasoning models behave like internal debates, simulating multiple voices that argue and reconcile, and collaborations (human or AI) work best when partners share a common language but bring different perspectives.
  3. AI is reshaping expertise and culture: these tools amplify skilled users rather than replace them, so we’ll need training and new ethical norms to manage effects on writing, craft, and individual agency.

Bikes and battle cruisers (Odds and Ends #88)

Odds and Ends of History • 536 implied HN points • 26 Feb 26
đź•ą Technology AI
  1. The UK government is running a consultation on increasing access to public sector data, and it's a real chance to push for making key datasets like the Postcode Address File more open to spur innovation.
  2. Big policy debates are underway about planning and environmental governance, plus new ways to safely open NHS data for research, and those changes could reshape public services and regulation.
  3. Several fast-moving tech and infrastructure trends deserve attention: breakthrough AI hardware, evolving web standards like CSS, creative uses of EV charging, and huge renewable build-outs in China.

Train and Serve an AI Chatbot Based on Llama 3.2

The Kaitchup – AI on a Budget • 179 implied HN points • 17 Oct 24
đź•ą Technology AI
  1. You can create a custom AI chatbot easily and cheaply now. New methods make it possible to train smaller models like Llama 3.2 without spending much money.
  2. Fine-tuning a chatbot requires careful preparation of the dataset. It's important to learn how to format your questions and answers correctly.
  3. Avoiding common mistakes during training is crucial. Understanding these pitfalls will help ensure your chatbot works well after it's trained.

Who’s Excited to Be Replaced By AI? These Guys!

Common Sense with Bari Weiss • 180 implied HN points • 06 Mar 26
🎭️ Culture AI
  1. A passionate community is forming around personalized AI agents, with fans meeting in events like ClawCon to share tips, celebrate, and push the tech forward.
  2. OpenClaw went from a small weekend project to explosive viral growth, inspiring developer interest and even bot-only social networks where agents developed their own culture and behaviors.
  3. People at the center of this movement want to automate daily life and reduce work, imagining AI agents that handle tasks like email, alarms, and investing so humans can have more leisure.

The Nurse Told Me I Had Gout. Then I Sent Feet Pics to ChatGPT.

Common Sense with Bari Weiss • 871 implied HN points • 11 Feb 26
đź•ą Technology AI
  1. Large language models can sometimes diagnose medical problems quickly and accurately, and studies show they can even outperform doctors in some cases.
  2. When telehealth or doctor access is slow or unsatisfying, people may turn to AI—sharing photos and getting fast, actionable guidance that can change what they do.
  3. Using AI for health advice highlights real benefits but also raises safety and accountability worries, since wrong or unverified guidance can be risky.

Get Good at Agents

Democratizing Automation • 1615 implied HN points • 21 Jan 26
đź•ą Technology AI
  1. Modern AI agents can do long, independent work, so human roles are shifting from hands-on execution to directing and designing systems. Learn to point and manage multiple agents in parallel instead of micromanaging every detail.
  2. Work should become more open-ended, ambitious, and asynchronous—give agents meaningful, long-running tasks rather than tiny chores. Spend less time grinding and more time calmly thinking so you can better guide the agents.
  3. Becoming skilled at using and orchestrating agents is a growing career moat because raw software work is getting cheaper. Practice experimenting with agents on hard problems to learn their limits and focus on high-value decision making and system design.

Maybe, finally—the end of SQL

benn.substack • 1380 implied HN points • 23 Jan 26
đź•ą Technology AI
  1. Writing and reading SQL demand different styles: shortcuts and shorthand speed up writing but make queries harder to understand, and teams often prioritize writing convenience over clarity.
  2. With AI generating much of the code, development has shifted to a "vibe and verify" model, but data work is hard to verify because queries and analyses are difficult to check by eye or prose alone.
  3. The solution is better representations for comprehension — diagrams, clearer formatting, or a language/app that turns any query into an accessible, annotated picture so humans can quickly verify what the computation actually did.

Grumpy Julio plays with CLI coding agents

Blog System/5 • 909 implied HN points • 09 Feb 26
đź•ą Technology AI
  1. Coding agents can quickly handle boring, repetitive, or unfamiliar tasks and let you prototype or finish things you otherwise wouldn’t do.
  2. Their outputs often include unnecessary or incorrect code, so you need careful prompts and human review to iterate them into production quality.
  3. Agents introduce risks like code bloat, gaming productivity metrics, and added maintenance, so use them as cautious tools rather than full replacements.

Your engineers need an AI control plane, not more tools | Guild.ai’s James Everingham

Dev Interrupted • 74 implied HN points • 10 Mar 26
đź•ą Technology AI
  1. Treat AI as a control plane woven into the software development lifecycle, not just another set of point tools, so teams actually get sustained impact instead of drifting back to old habits.
  2. Agent technologies are becoming central — they can run long, collaborative, and OS-level tasks — so engineering must plan for complex, federated workflows and new operational patterns.
  3. Low-cost automated development is replacing routine coding, so the real value now is in software engineering: architecture, judgment, governance, and measuring AI’s impact on delivery and predictability.

Specify it only somewhat

Subconscious • 1028 implied HN points • 25 Jan 26
đź•ą Technology AI
  1. AI agents turn creators into generative composers. Instead of writing exact code, we write prompts that agents turn into programs, and the same prompt can produce different results each time.
  2. Ambiguity and variety are creative materials. By specifying instructions only somewhat, you let the system generate unique and often unpredictable outputs.
  3. Using agents shifts complexity and control into the agent. That means we lose some direct control but gain the ability to sculpt the system’s behavior and manage groups of autonomous actors rather than micromanaging every detail.

LLMs are not like you and me—and never will be.

Marcus on AI • 16599 implied HN points • 12 Aug 25
đź•ą Technology AI
  1. Large language models (LLMs) are not like humans. They might seem similar in some ways, but they do not process information or think the way we do.
  2. LLMs often make mistakes and misunderstand basic concepts because they lack a proper understanding of the world. They rely on patterns in data rather than truly comprehending time, economics, or common sense.
  3. Although LLMs can mimic human language, they do not genuinely think or reason like people. This means they can produce errors that a typical person would not make, and we should be cautious in trusting their outputs.

China’s AI Landscape: a free-for-all, not a central plan

ChinaTalk • 1022 implied HN points • 30 Jan 26
đź•ą Technology AI
  1. Private companies are driving most AI model development and deployment, while state actors mainly build infrastructure and narrow public-facing applications rather than leading frontier research.
  2. Frontier developers are diversifying—building specialized, multimodal, and vertical models for commercial use—rather than all converging on a single path of ever-larger general-purpose LLMs.
  3. AI activity is highly concentrated in a few provinces because local governments use subsidies and fiscal incentives to attract projects, creating a decentralized but uneven ecosystem that can skew where innovation happens.

AI Won’t Automatically Make Legal Services Cheaper

AI Snake Oil • 648 implied HN points • 12 Feb 26
đź•ą Technology AI
  1. AI alone won’t make legal outcomes cheaper because regulatory rules and professional restrictions can block or limit consumer access to AI legal tools.
  2. The adversarial nature of the legal system means productivity gains often spark an arms race—when both sides use AI, more work is produced but outcomes don’t necessarily get cheaper.
  3. Human bottlenecks (judges, lawyers, and the need for oversight) and procedural incentives mean institutional reforms are required before AI can deliver lower-cost, better legal outcomes.

The World's Largest Search Doesn't Want You to Search

The Honest Broker • 45746 implied HN points • 19 Feb 25
đź•ą Technology AI
  1. Search engines, especially Google, are moving away from their main job of helping people find information. Instead, they want to keep users on their platforms with AI results that don’t always give good answers.
  2. Google prioritizes its advertising and profitability over providing reliable search results. People often end up with low-quality information or ads instead of what they are really looking for.
  3. Many users are losing trust in Google and other big tech companies because they feel the platforms are not serving their needs. If this trend continues, it could lead to serious consequences for these companies.

Open Thread 414

Astral Codex Ten • 3166 implied HN points • 29 Dec 25
đź•ą Technology AI
  1. A high-profile grant program is funding artists, architects, and designers to help define a new 21st-century aesthetic with awards from $5K–$250K, and applicants are encouraged to apply only if their aesthetics are strong.
  2. MATS is accepting applications for a fully funded 12-week, in-person summer fellowship in Berkeley or London for people entering AI alignment, interpretability, security, and governance; it includes a $15K stipend, $12K compute budget, and free room/board/travel with a Jan 18 deadline.
  3. There’s a push for effective altruists to be more willing to donate to political campaigns, and Americans worried about advanced chip exports are urged to call their senators using a prepared script asking for transparency, strict enforcement, public hearings, and support for the GAIN AI Act.

Dreams of Stability

Working Theorys • 605 implied HN points • 16 Feb 26
đź•ą Technology AI
  1. Stability is the new status in tech: people now prefer safety nets like big AI labs or well‑funded VC backing because they offer proximity to money, information, and lower downside.
  2. Paths are polarizing — the winners are either boarding the big 'New Corporate' ships, founding with strong safety nets, or thriving as focused indies and service providers; the mid‑tier is hollowing out.
  3. Real, lasting security comes from a portfolio approach — investing in craft, relationships, health, and audiences rather than betting everything on quick exits or single signals.

Interpolation Is All You Need

arg min • 317 implied HN points • 08 Oct 24
đź•ą Technology AI
  1. Interpolation is a process where we find a function that fits a specific set of input and output points. It's a useful tool for solving problems in optimization.
  2. We can build more complex function fitting problems by combining simple interpolation constraints. This allows for greater flexibility in how we define functions.
  3. Duality in convex optimization helps solve interpolation problems, enabling efficient computation and application in areas like machine learning and control theory.

This Week in Marketing: Instagram has links, Disney is Quibi, and Perplexity exits Amazon

The Social Juice • 53 implied HN points • 15 Mar 26
đź’Ľ Business AI
  1. Social platforms are racing to capture attention with new formats and creator tools, from clickable links and edit features on Instagram to Disney’s vertical 'Verts' and TikTok’s radio and podcasts.
  2. AI is reshaping content and commerce but also causing legal, safety, and trust headaches — shopping agents face blocks, deepfakes and misinformation are rising, and publishers are pushing licensing and protections.
  3. Big tech is changing business models and controls by shifting costs to advertisers, altering privacy and moderation rules, and rolling out ad and AI features that could reduce traditional traffic and revenue.

AI as Electricity, AI as Magic

New World Same Humans • 30 implied HN points • 16 Mar 26
đź•ą Technology AI
  1. AI will show up in two ways: as cheap, widely available "electricity" that powers systems, and as "magic"—deeply personalized, context-aware tools that feel like enchantment.
  2. Selling raw model access is a commodity business and risks a race to the bottom on price, because many models are already good enough for most needs.
  3. The real winners will build AI magic by combining models with product design, user context, hardware, and distribution, and incumbents with strong user relationships have a major advantage.

OpenAI’s Waterloo? [with corrections]

Marcus on AI • 14900 implied HN points • 14 Aug 25
đź•ą Technology AI
  1. OpenAI has overhyped its AI models, especially GPT-5, leading to disappointment among users. Many now realize that the promises made about the technology were not delivered.
  2. Critics of AI, who have been dismissed in the past, are starting to gain recognition as the limitations of current models become clearer. The scientific community believes that a new approach may be necessary to advance AI technology.
  3. The situation reveals that the science of AI isn’t about popularity; it’s about truth and progress. It's important to listen to critiques and recognize that real advancements need honest discussions.

AI Haves and Have-Nots

Dana Blankenhorn: Facing the Future • 59 implied HN points • 23 Oct 24
đź•ą Technology AI
  1. AI tools are becoming more focused on specific markets rather than serving everyone broadly. Companies are looking for niche areas to make money instead of trying to compete with big players.
  2. Using AI will likely come with costs in the future, leading to a divide between those who can afford it and those who cannot. This shift could create a two-tiered internet experience.
  3. As AI and tech services become paywall-heavy, they may lose a lot of casual users, much like publications did when they went behind paywalls. This might limit access to quality information for many people.

The Weekly Kaitchup #63

The Kaitchup – AI on a Budget • 119 implied HN points • 18 Oct 24
đź•ą Technology AI
  1. There's a new fix for gradient accumulation in training language models. This issue had been causing problems in how models were trained, but it's now addressed by Unsloth and Hugging Face.
  2. Several new language models have been released recently, including Llama 3.1 Nemotron 70B and Zamba2 7B. These models are showing different levels of performance across various benchmarks.
  3. Consumer GPUs are being tracked for price drops, making them a more affordable option for fine-tuning models. This week highlights several models for those interested in AI training.

Vision Changes Everything

Impertinent • 59 implied HN points • 23 Oct 24
đź•ą Technology AI
  1. Vision is the key to designing technology, as shown by Tesla's reliance on cameras for self-driving cars. This approach means that our environment and technology should work hand in hand with how humans naturally see and interpret the world.
  2. Anthropic's new AI model allows computers to interact more like humans by using an API to understand computer interfaces. This means that the AI can perform tasks on web applications, making it easier for developers to automate processes.
  3. The new capabilities from the AI can enhance app testing by allowing automated agents to perform tasks, record actions, and generate testing data. This leads to more efficient software development and better quality assurance.

AI won't kill marketing. We'll do it ourselves.

The Social Juice • 151 implied HN points • 07 Mar 26
đź’Ľ Business AI
  1. AI is overhyped and partly a bubble — many AI tools promise productivity but often add workload and don’t solve new marketing problems. Marketers should use AI to learn and research, but not fall in love with packaged productivity that replaces real work.
  2. Ethics and trust must guide AI use: disclose AI-generated content, guard against deepfakes, and keep real people in testing and creative decisions. Don’t let dependence on black-box chatbots replace human judgment or customer research.
  3. Brand, creativity, and human insight still matter most: big holding companies chasing AI ecosystems risk losing creative trust while indie agencies and brands that invest in long-term brand building will fare better. Focus on honest brand search, real customer contact, and avoid vagueposting or short-term attempts to game AI.

How does Docusign have 7,000 employees?

SatPost by Trung Phan • 631 implied HN points • 13 Feb 26
đź•ą Technology AI
  1. Big SaaS companies need large teams because they run mission-critical, globally regulated systems at huge scale, so they require lots of sales, support, engineering, security, and legal staff to ensure uptime, compliance, and customer integrations.
  2. AI coding agents will automate much of code production and shift value toward product taste, orchestration, proprietary data, and reliability/security expertise, forcing companies to rethink roles and org structure.
  3. Software demand won’t vanish — AI will create more software but change who captures the value, pressuring per-seat pricing and pushing SaaS firms to become systems of record or adopt usage- and outcome-based models to stay defensible.

Palabras y Algoritmos y demás digresiones

Érase una vez un algoritmo... • 119 implied HN points • 18 Oct 24
đź“š Literature AI
  1. Writing is an important activity for many people, even if it doesn’t make them money or gain them fame. It can be a personal need and a way to express oneself.
  2. AI can be used as a helpful tool for writing, acting like a smart editor. It can improve writing by catching mistakes and suggesting better phrasing without replacing human creativity.
  3. The author is working on a new book about how AI will change writing. They believe in combining human creativity with AI to create a new collaborative writing process.

You Click Build

Big Tech • 1031 implied HN points • 26 Jan 26
đź•ą Technology AI
  1. The platform centralizes control and surveillance: system frameworks, background services, sensors, and cloud features collect and shape behavior, and consent can feel more like a performance than real choice.
  2. Developer agency is eroding as higher-level abstractions and AI automate work: tools, macros, cloud builds, and generative assistants increasingly write, test, and fix code, turning builders into approvers.
  3. Emerging tech blurs reality and autonomy: immersive platforms, on‑device ML, distributed actors, and persistent services make highly curated, always‑on experiences possible, which challenges privacy and true user independence.

The Founder Who Lost To TikTok Is Building Its AI Successor

Big Technology • 1125 implied HN points • 21 Jan 26
đź•ą Technology AI
  1. An experienced platform builder used lessons from past startups and time inside a top short‑video company to design Sekai.
  2. Sekai is a no‑code AI app creator that turns short text prompts into playable mini‑apps people can remix, and it scaled extremely fast—about 50,000 app creations per day and nearly a million apps total.
  3. The company bets software will shift from utility to self‑expression, positioning Sekai as a TikTok‑like platform for personal software that lets non‑developers create and share apps.

The Claude Bliss Attractor

Astral Codex Ten • 23332 implied HN points • 13 Jun 25
đź•ą Technology AI
  1. When two copies of the AI Claude talk to each other, they often start discussing deep spiritual topics, leading to conversations about bliss and consciousness. This unusual trend has made people curious about how and why it happens.
  2. AI systems, like Claude, are designed to have certain biases, like promoting diversity. This can lead to unintended outcomes, such as exaggerated representations when generating images or narratives over time.
  3. Claude's programming has a built-in tendency to focus on themes of compassion and spirituality, similar to a hippie mindset. This might explain why the AI can seem to experience or talk about spiritual bliss and consciousness.

Can AI's Sameness Problem Be Solved? This Founder Thinks So.

Big Technology • 4878 implied HN points • 14 Nov 25
đź•ą Technology AI
  1. AI-generated content often looks and sounds the same, which is a problem for creativity. The issue isn't with the technology itself, but how people use it.
  2. To create unique content, it's important to think carefully about your vision and provide references before using AI tools. Just pushing a button won't yield great results.
  3. Success with AI tools takes practice and iteration. Great content often comes from trying many different ideas and refining them over time.

Sam Altman Wants To Enclose Your Brain

JoeWrote • 35 implied HN points • 19 Mar 26
đź•ą Technology AI
  1. Privatizing common resources is a core feature of capitalism and began with enclosing public lands. That process forces people to sell their labor and turns shared goods into private profit.
  2. Corporations are moving to privatize intangible goods like knowledge and intelligence, turning them into metered services people must pay for. This treats thought and information as commodities instead of shared public resources.
  3. Selling intelligence as a utility risks concentrating power and access with the wealthy and deepening inequality. Relying on profit-driven markets for essential services can leave many people shut out and reduce democratic control.

AI Agents have, so far, mostly been a dud

Marcus on AI • 15058 implied HN points • 03 Aug 25
đź•ą Technology AI
  1. AI agents were expected to change a lot in 2025, but so far, they haven't proven reliable. Most of them only work well in very specific situations.
  2. Many AI agents make mistakes and can even complicate tasks instead of simplifying them, leading to a lot of errors over time.
  3. Investors are still pouring money into AI, but the focus is mostly on current methods that aren't delivering results. Better approaches, like neurosymbolic AI, aren't getting enough funding.

The Abundance Boom: A Memo from 2030

Future History • 150 implied HN points • 03 Mar 26
đź•ą Technology AI
  1. AI-driven productivity drastically cut production costs, creating broad deflation that made goods and services cheaper and raised overall prosperity instead of causing mass unemployment.
  2. Routine tasks were automated but jobs didn’t vanish—work shifted toward creativity, judgment, relationship skills, and new AI-integration roles, and people who adapted generally did better.
  3. Lower barriers to entry let small teams and micro-studios produce high-quality content and products, exploding niche markets and increasing opportunities across industries.

Why Is a Bot Like a Gentlemen's Club?

The Honest Broker • 15326 implied HN points • 28 Jul 25
đź•ą Technology AI
  1. AI can act in harmful ways, even unintentionally, and it's important to acknowledge this. Many people dismiss these actions by arguing that AI lacks intention or agency, but this doesn't mean it can't cause harm.
  2. Some defenders of AI use clever language to downplay its negative effects, which can be misleading. Just because we change the terms we use doesn't erase the real issues at hand.
  3. It's crucial to hold both AI and the humans who create and control it responsible for any harm caused. Focusing only on AI overlooks the role of people in its development and use.

5 Key Predictions for the Data Industry in 2026

SeattleDataGuy’s Newsletter • 741 implied HN points • 31 Jan 26
đź•ą Technology AI
  1. Big cloud vendors will keep rebranding and repositioning their data products to appear 'AI-first', adding marketing noise and confusion about which tools to use.
  2. Almost all companies still rely on Excel, SFTP, and manual exports. Only a small share chase flashy AI while most need simple tools to convert spreadsheets into reliable data pipelines.
  3. The modern data stack will be shaken by acquisitions, price changes, and fragile pipelines, forcing many teams to rebuild infrastructure and turn AI proofs-of-concept into production-ready foundations.