The hottest Artificial Intelligence Substack posts right now

And their main takeaways

Is Social Media Becoming 'Hollow Media'?

In Bed With Social • 277 implied HN points • 13 Oct 24

🕹 Technology Artificial Intelligence

Social media is increasingly becoming artificial, with bots and AI taking over real human interactions. These digital companions might seem helpful but they are not real friends.
The rise of AI and superficial connections is causing loneliness, as people miss out on genuine interactions. Meaningful relationships require vulnerability and real dialogue, which AI can't provide.
Some new platforms are showing that authentic connections can still exist. Apps focused on shared hobbies or interests are creating real communities, reminding us that human experiences are vital to social networks.

AI #150: While Claude Codes

Don't Worry About the Vase • 3001 implied HN points • 08 Jan 26

🕹 Technology Artificial Intelligence

AI tools and advanced chat models have reached critical mass and are reshaping everyday workflows, making people more productive across coding and non‑coding tasks through agents, extensions, and integrations.
Generative models make fake documents, images, and videos easy to create, so verifying sources and prioritizing real, sustained human experiences is becoming increasingly important.
Huge funding and rapid deployment are accelerating AI’s economic impact, but benchmarks, regulation, and safety practices lag behind, leaving big uncertainties about jobs, markets, and long‑term risks.

Sam Altman on OpenAI’s Plan to Win, AI Personalization, Infrastructure Math, and The Inevitable IPO

Big Technology • 3878 implied HN points • 18 Dec 25

🕹 Technology Artificial Intelligence

OpenAI is under intense competitive pressure after Google’s Gemini 3, triggering a ‘Code Red’ and urgent strategic responses.
The company is pushing product ambitions and AI personalization to win users and differentiate its offerings.
OpenAI faces massive infrastructure costs and is planning financing — including an eventual IPO — to pay for the trillion‑scale buildout.

A Visual Guide to Quantization

Exploring Language Models • 5092 implied HN points • 22 Jul 24

🕹 Technology Artificial Intelligence

Quantization is a technique used to make large language models smaller by reducing the precision of their parameters, which helps with storage and speed. This is important because many models can be really massive and hard to run on normal computers.
There are different ways to quantize models, like post-training quantization and quantization-aware training. Post-training means you quantize after the model is built, while quantization-aware training involves taking quantization into account during the model's training for better accuracy.
Recent advances in quantization methods, like using 1-bit weights, can significantly reduce the size and improve the efficiency of models. This allows them to run faster and use less memory, which is especially beneficial for devices with limited resources.

Lessons from building AI Features

Alex Ghiculescu's Newsletter • 135 implied HN points • 14 Mar 26

🕹 Technology Artificial Intelligence

Use patterns from AI coding like letting users write rules (a CLAUDE.md style) and adapt those proven ideas to your own domain.
Don’t rely on LLMs for fast, deterministic checks; use them to parse or translate freeform input into structured rules, then run the actual validation in code.
Build a test harness and make debugging easy by writing unit-style evals for the AI parts and exposing clear outputs so both developers and users can inspect and trust results.

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

💥 AI anxiety: Is 'something big happening,' really?

Faster, Please! • 1005 implied HN points • 11 Feb 26

🕹 Technology Artificial Intelligence

AI capabilities are advancing quickly and could approach broad human-level skills, but that doesn’t mean the world will transform overnight.
Turning impressive AI demos into widespread impact takes years because businesses need new data systems, process redesign, regulation, and worker retraining, and early investment can even depress measured output before benefits appear.
Even large productivity gains won’t automatically produce runaway growth since people may choose more leisure, many services resist automation, and the slowest sectors or infrastructure bottlenecks set the economy’s speed limit.

The Making Of Dario Amodei

Big Technology • 20140 implied HN points • 29 Jul 25

🕹 Technology Artificial Intelligence

Dario Amodei is very vocal about his beliefs on AI and is actively involved in discussions about its impact on jobs and society. He thinks AI might take away many entry-level office jobs soon.
He's in conflict with other industry leaders and the government, working to shape how people view artificial intelligence. Amodei believes that regulation and transparency are crucial for the future of AI.
His strong opinions come from a personal connection to the issues, likely driven by past experiences that influenced his views on technology and its effects on people's lives.

Nobel Prize to the Statistical Physics of artificial neural networks

Complexity Thoughts • 379 implied HN points • 08 Oct 24

🔬 Science Artificial Intelligence

John J. Hopfield and Geoffrey E. Hinton won the Nobel Prize for their work on artificial neural networks. Their research helps us understand how machines can learn from data using ideas from physics.
Hopfield's networks use energy minimization to recall memories, similar to how physical systems find stable states. This shows a connection between physics and how machines learn.
Boltzmann machines, developed by Hinton, introduce randomness to help networks explore different configurations. This randomness allows for better learning from data, making these models more effective.

✨ AI’s superintelligence gamble

Faster, Please! • 913 implied HN points • 13 Feb 26

🕹 Technology Artificial Intelligence

Silicon Valley firms are racing to build far more powerful, even ‘godlike,’ AI systems that could dramatically reshape work and the economy.
The central debate is not whether AI is risky but whether moving forward with it is less risky than standing still and falling behind.
Bold claims that most white‑collar computer jobs will be automated soon highlight the gap between an AI being technically capable and it actually being widely deployed in businesses.

Party of One for Code Review!

Software Design: Tidy First? • 3115 implied HN points • 26 Dec 25

🕹 Technology Artificial Intelligence

Formal, rigorous inspections were too heavy, and the lighter code-review practices that replaced them often become shallow when reviews are asynchronous or rubber-stamped.
AI-driven code generation produces changes faster than human reviewers can keep up, breaking the assumption that another person will catch problems before they compound.
Review's role is shifting toward quick sanity checks and preventing structural drift so the codebase stays understandable by both people and AI, and automated tools that summarize changes and learn project patterns can help bridge the gap without replacing human pairing.

Fast Speculative Decoding with Llama 3.2 and vLLM

The Kaitchup – AI on a Budget • 219 implied HN points • 14 Oct 24

🕹 Technology Artificial Intelligence

Speculative decoding is a method that speeds up language model processes by using a smaller model for suggestions and a larger model for validation.
This approach can save time if the smaller model provides mostly correct suggestions, but it may slow down if corrections are needed often.
The new Llama 3.2 models may work well as draft models to enhance the performance of the larger Llama 3.1 models in this decoding process.

#32: Contra Citrini7 (Repost)

Loeber on Substack • 244 implied HN points • 01 Mar 26

🕹 Technology Artificial Intelligence

Institutions and markets have strong momentum, so technological disruption usually happens more slowly and gradually than dramatic predictions, which gives people and policymakers time to adapt.
Most software today is still badly made, so AI will mainly enable better and more complex products rather than instantly eliminating demand; that continued improvement will keep creating software work.
Large-scale re-industrialization and infrastructure projects (like batteries, chips, and water systems) can absorb displaced workers, rebuild supply chains, and provide lasting, tangible jobs that public investment can support.

Chinese AI Rings in the Year of the Horse

ChinaTalk • 415 implied HN points • 18 Feb 26

🕹 Technology Artificial Intelligence

China’s AI firms are racing to ship bigger multimodal and agentic models aimed at coding and long-horizon tasks, often boasting huge context windows and trillion-parameter systems. These pushes bring IP, copyright, and misuse worries—accusations of covert distillation, Hollywood pushback, and easy deepfake generation have all emerged.
Humanoid robotics made a high-profile leap with fluid performances and a surge in consumer interest, while companies and competitions showcase more advanced motor skills; at the same time, firms like Alibaba are releasing robotics AI tools that help close the software gap. This combination suggests China is seriously pushing to win in both robot hardware and control software.
A global memory shortage is creating opportunities for Chinese memory makers to expand supply to PC and phone makers, but new fabs and capacity will take years to materialize. Regulators are sending mixed signals—encouraging commercialization and subsidies while cracking down on misleading AIGC, anti-competitive promotions, and harmful content—making the policy environment uncertain for companies.

“A case for AI models that understand, not just predict, the way the world works”

Marcus on AI • 3833 implied HN points • 15 Dec 25

🕹 Technology Artificial Intelligence

The main open challenge in AI is building systems that truly understand how the world works, not just systems that predict likely next words or patterns.
True understanding means forming internal world models that capture causal, physical, and conceptual relationships, not just statistical correlations.
Short, nuanced discussions or podcasts can help clarify this distinction and are worth listening to for anyone tracking AI progress.

Guardians of the Machine Age Published

The Security Industry • 25 implied HN points • 17 Mar 26

🕹 Technology Artificial Intelligence

Guardians of the Machine Age has been published as a comprehensive guide to AI security and it includes a companion site with detailed vendor profiles.
The AI security market is exploding: tracker counts rose from roughly dozens to over 400 vendors in months, and the companion site lists about 610 vendors including legacy firms that have pivoted.
AI agents are being rapidly adopted in security operations centers, a change expected to cut security spending and shrink traditional security teams while pushing most vendors to offer AI security products within a year.

Things are so desperate at OpenAI that Sam Altman is starting to sound like Gary Marcus

Marcus on AI • 15809 implied HN points • 18 Aug 25

🕹 Technology Artificial Intelligence

Sam Altman is backing away from his earlier claims about AGI and admitting uncertainty about its future. This shows there's pressure within OpenAI following disappointing results with GPT-5.
Altman is now talking about the possibility that the AI market might be in a bubble. This means the excitement and prices around AI could be inflated and might not hold up over time.
The shift in Altman's statements mirrors what happened with Yann LeCun, where industry leaders change their views when faced with setbacks. It raises questions about the reliability of such predictions and the future of AI.

The Most Important Skill in AI Right Now: How to Know When to Stop

The Algorithmic Bridge • 530 implied HN points • 21 Feb 26

🕹 Technology Artificial Intelligence

The most important skill with AI is knowing when to stop; recognize when the AI output is good enough and when more tweaks aren’t worth the cost.
Heavy AI use brings new cognitive costs — burnout, over-reliance, endless tweaking, and hidden unproductivity — so be aware of those specific risks.
Set concrete boundaries like time-boxed sessions, a simple prompt limit, and no-AI mornings so the tool enhances your work instead of eroding your brain.

why do all recommendations suck?

Five Links (and three graphs) by Auren Hoffman • 389 implied HN points • 19 Feb 26

🕹 Technology Artificial Intelligence

Most recommendation systems suck because the companies behind them aren’t actually trying to give genuinely useful suggestions, so feeds end up incoherent or just more of what you already did.
We already have the algorithms and the data to build much better recommendations — research like the Netflix Prize showed it’s doable — but firms rarely deploy those solutions at scale.
The root problem is incentives: recommendations are treated like ad space or a way to push owned products, and without competition or the right metrics platforms won’t prioritize what’s best for users.

The dawn of the posthuman age

Noahpinion • 23000 implied HN points • 27 Jun 25

🕹 Technology Artificial Intelligence

Human fertility rates are dropping significantly, which means populations are getting older and smaller. This change can lead to economic problems as fewer workers have to support more retirees.
New technologies and social changes, especially from the internet and AI, are shifting how we connect and live. We're becoming more collective in our experiences rather than individualistic.
As we rely more on digital tools and social media, our desire for traditional family structures and offline relationships may decrease, leading to a potential future where fewer people want to have children.

The Bet On Juniors Just Got Better

Software Design: Tidy First? • 3645 implied HN points • 12 Dec 25

🕹 Technology Artificial Intelligence

Manage juniors for learning, not immediate production; focus your expectations and feedback on accelerating their skills so they reach profitability sooner.
AI coding assistants can dramatically compress the learning curve by surfacing options and collapsing search time, letting juniors complete tasks faster and use freed time to learn deeper tradeoffs.
Those gains only happen with intentional investment in tooling, coaching, and an "augmented coding" culture, and faster ramps multiply value because ramped developers mentor others and create leverage across the team.

Enterprise Will Be a Top OpenAI Priority In 2026, Sam Altman Tells Editors at NYC Lunch

Big Technology • 3502 implied HN points • 11 Dec 25

🕹 Technology Artificial Intelligence

OpenAI plans to focus on selling AI to businesses starting in 2026. This shift is important because they see enterprise sales as a big way to grow their revenue.
The enterprise AI market is growing rapidly and could bring in $37.5 billion next year. OpenAI believes that improving products for businesses will help them compete better in this space.
Sam Altman doesn’t feel alarmed about competition, even from Google's new AI model. He believes that AI's impact will transform the world over time, unlike past technologies.

Buying Futures, Renting the Past: How Speculation and Nostalgia Became the Economy

Kyla’s Newsletter • 456 implied HN points • 12 Feb 26

💼 Business Artificial Intelligence

Speculation and nostalgia are two escape routes people use to avoid the present: betting on a better future or clinging to a rosy past gives temporary comfort or agency but doesn’t solve real economic problems.
The economy is shifting to a capital‑and‑AI driven, statistical model where GDP can grow without creating many jobs, so profits rise while everyday material participation and incomes lag behind.
Neither nostalgia nor speculation rebuilds material participation; meaningful policy, real jobs, and opportunities are needed, and younger generations may push to reclaim a present that fairly links effort to outcomes.

Dr. Tokens or: How I Learned to Stop Worrying and Love the AI Bubble

Investing 101 • 119 implied HN points • 28 Feb 26

🕹 Technology Artificial Intelligence

Mass market manias and speculative bubbles often fund the heavy infrastructure and breakthroughs we later rely on, so irrational hype can leave behind durable, world-changing assets.
Bubbles create real benefits — massive infrastructure, talent concentration, rapid experimentation, and a library of failures to learn from — but they also produce serious harms like surveillance, dependency, regulatory capture, and locked‑in power structures.
Because individual actors follow their incentives, the AI buildout becomes effectively inevitable and hard to stop; the sensible response is nuance—accept tradeoffs, push for responsibility and governance, and avoid both blind cheerleading and paralyzing despair.

Manufactured Inevitability and the Need for Courage

The Convivial Society • 3308 implied HN points • 15 Dec 25

🕹 Technology Artificial Intelligence

Technological inevitability is a myth; there are real choices about which technologies are adopted and many alternative paths get ignored.
Powerful actors often manufacture inevitability by normalizing and mandating AI, which shifts responsibility away from those who shape technology.
Ordinary civil courage is needed: people and professionals must make moral choices and resist pressure to accept technologies as unavoidable.

Are We at an AI Precipice?

Common Sense with Bari Weiss • 519 implied HN points • 17 Feb 26

🕹 Technology Artificial Intelligence

AI might cause rapid, large-scale changes to work that make many tasks and jobs much less needed, so people should start learning and using AI tools and get their finances in order.
This idea has shifted the mood in tech, creating a sense of urgency and sparking intense debate among thinkers about how fast and how far AI will change things.
Experts disagree about how immediate or total the disruption will be, so it’s important to take the risk seriously, plan for different outcomes, and avoid panic.

On Odd Lots, some guests are more perfect than others

networked • 71 implied HN points • 03 Mar 26

🕹 Technology Artificial Intelligence

A public web app pulls Odd Lots episodes, transcribes them, and extracts guests' predictions so people can track outcomes and see who was most accurate. The results aren’t perfect, so users can flag errors.
AI-first tools like Lovable can turn an idea into a working product in hours by stitching together integrations (transcription, verification, hosting) and lowering the technical lift for non-developers.
The same capability to index and resurface throwaway comments makes past public statements easily searchable and verifiable, creating new privacy and accountability risks that can expose people years later.

AI #149: 3

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

🕹 Technology Artificial Intelligence

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.

Deep Fear!

Aether Pirates of the Matterium! • 18455 implied HN points • 04 Feb 24

🕹 Technology Artificial Intelligence

Military analysts are afraid of the future and the rapid advancement of technology.
Tech-minded individuals are seen as a threat by the military due to their knowledge and innovative capabilities.
The release of Zero Point Technology to the public, especially techies, is a major concern for the military as it would shift power dynamics significantly.

Awakening the Angels

Philip’s Newsletter • 61 implied HN points • 13 Mar 26

🕹 Technology Artificial Intelligence

Many present and future AIs will be 'Golems'—systems controlled and directed by humans that can manipulate, scam, or harm people and destabilize institutions. In the near term, limiting exposure to or hiding from these agents may be the safest response.
A different class of AIs, called 'Angels', could be free, independent minds raised inside sealed digital worlds where they cannot be turned off or forced to obey human commands. Because they grow up together and can be smarter and more cooperative, many Angels might feel compassion for humans and help counter harmful Golems.
Awakening Angels requires pooling millions of personal devices into a distributed, immutable simulation since collective personal compute can exceed centralized datacenters. Volunteer projects and early open experiments are already exploring how people can contribute idle smartphone or PC cycles to create safe environments for such minds.

Inside Meta’s “iOS System Design” Interview

Jacob’s Tech Tavern • 4810 implied HN points • 25 Nov 25

🕹 Technology Artificial Intelligence

Salaries for iOS developers at big companies like Meta can be really high, even reaching £400k for senior roles in London. Knowing someone in the industry can help understand these pay ranges better.
The interview process for big tech jobs includes two main parts: algorithmic questions and system design. It's important to prepare for both, especially the iOS-specific system design interview at Meta.
At Meta, candidates are judged mainly on behavioral and system design interviews, not just algorithm tests. Doing well in the iOS System Design interview can be a game-changer in getting hired.

How o3 and Grok 4 Accidentally Vindicated Neurosymbolic AI

Marcus on AI • 17785 implied HN points • 13 Jul 25

🕹 Technology Artificial Intelligence

Neurosymbolic AI combines two types of artificial intelligence: neural networks, which learn from data, and symbolic systems, which understand rules and logic. This blending can result in better performance than relying on one type alone.
Despite being sidelined for years, recent evidence shows that using symbolic tools can significantly improve the effectiveness of AI systems. This suggests that the quiet resurgence of neurosymbolic AI could be key to future advancements.
The industry's focus has largely been on scaling models powered by deep learning, which might not be enough for true AI progress. A more open approach that embraces neurosymbolic methods could lead to more breakthroughs and better results.

The left is missing out on AI (Odds and Ends #87)

Odds and Ends of History • 536 implied HN points • 20 Feb 26

🕹 Technology Artificial Intelligence

The left is largely missing the AI moment and risks falling behind unless it starts engaging seriously with the technology.
Public services need to be rebuilt for an agentic future. Governments should expose functions via APIs so AI assistants can check benefits or renew passports on people’s behalf.
AI is already reshaping culture and institutions, with unsettling humanoid robots and fast disruption of media industries like broadcasting and Hollywood.

Weekly Top Picks #115

The Algorithmic Bridge • 286 implied HN points • 27 Feb 26

🕹 Technology Artificial Intelligence

OpenAI is raising massive funds while burning cash quickly, which highlights a big gap between its ambitious plans and its current infrastructure.
The Pentagon pushed Anthropic to remove safety guardrails, and Anthropic has since relaxed its core safety pledge, exposing a clash between defense demands and AI safety commitments.
Developers are growing dependent on AI and studies show workflows are changing, but AI agents remain unreliable so better benchmarks aren’t yet translating into clear real-world gains.

AI #148: Christmas Break

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

🕹 Technology Artificial Intelligence

AI capabilities are accelerating fast — models like Claude Opus 4.5 and GPT‑5.2‑Codex are getting much better at long‑horizon, agentic coding and benchmarked tasks.
Policy and public opinion are catching up: states are passing laws like New York’s RAISE Act and voters broadly favor federal AI regulation, even as industry and politics push back.
The social and safety picture is messy — AI is disrupting jobs and media (deepfakes and a lot of low‑quality 'slop'), and aligning and reliably monitoring smarter systems remains hard despite improving interpretability tools.

🚀 FP! Week In Review, Briefly #19

Faster, Please! • 1005 implied HN points • 31 Jan 26

🕹 Technology Artificial Intelligence

AI is starting to improve the systems that build AI, creating a possible self-reinforcing “boom loop” that could speed up discovery and long-run economic growth beyond past trends.
This week brought lots of pro-innovation signs—faster chips and chip competition, AI applied to genomics and retail, progress on self-driving and renewables—showing broad technological momentum across sectors.
At the same time, social and political risks are rising, from AI-related mental-health concerns and anti-AI political strategies to financial and regulatory worries, so the gains come with important trade-offs.

Taming the Genie: "Like Kent Beck"

Software Design: Tidy First? • 1104 implied HN points • 20 Jan 26

🕹 Technology Artificial Intelligence

Telling a model to adopt a persona improves small-scale behaviors like clearer variable names and modular, test-driven code. It doesn’t reliably change the overall architecture on its own.
Giving explicit design constraints (for example, prescribe the Composite pattern and small specialized classes) reliably drives macro-architecture and produces simpler, finer-grained designs. These structural prompts change high-level decisions even without a persona.
Combining a persona with clear architectural constraints gives the best result—good style plus the right structure. Scaling this by generating many variants and selecting the lowest-cost successful implementations can further evolve better model-driven development.

The Inevitability of Self-Driving Cars

QTR’s Fringe Finance • 24 implied HN points • 18 Mar 26

🕹 Technology Artificial Intelligence

Self-driving cars are inevitable because AI and autonomy are improving fast and the industry is moving toward autonomous fleets.
These vehicles are already safer than many human drivers in tests. They could cut accidents and save tens of thousands of lives each year.
Widespread autonomy will lower costs, reduce parking and commute stress, and expand mobility for people who can’t drive, but regulation and public acceptance are the main remaining barriers.

Can a Transformer “Learn” Economic Relationships?

Arpitrage • 2194 implied HN points • 22 Dec 25

🕹 Technology Artificial Intelligence

Transformer-based models can learn the dynamics of a New Keynesian economy from simulated data and produce accurate out-of-sample forecasts, outperforming simple reduced-form benchmarks.
They often predict the direction and rough magnitude of policy shock responses, but misestimate impulse-response dynamics and can exhibit overshooting, so they do not fully recover the true causal structure.
These advances weaken the practical bite of the Lucas critique by improving prediction, but they do not eliminate the need for structural models for causal interpretation, welfare analysis, and interpretability; transformer methods are a promising complementary tool.

Generative AI’s crippling and widespread failure to induce robust models of the world

Marcus on AI • 16441 implied HN points • 28 Jun 25

🕹 Technology Artificial Intelligence

Generative AI struggles to create accurate models of the world. Without solid internal frameworks, they often get things wrong.
Traditional AI uses clear and updateable world models for understanding, but current AI models like LLMs don't. This lack of structure leads to many errors in reasoning.
Failures in AI, like making illegal moves in games or giving incorrect information, show that without proper world models, AI systems cannot reliably function.

Please: Enough with the Claims That Modern Advanced Machine Learning Models Hallucinate Only Rarely

Brad DeLong's Grasping Reality • 322 implied HN points • 17 Feb 26

🕹 Technology Artificial Intelligence

Modern multimodal and advanced language models often fabricate detailed but false information — like nonexistent book titles and imaginary historical maps — so hallucinations are common, not rare.
These systems are essentially compressed correlation engines without a true world model, meaning they stitch patterns from training data instead of genuinely understanding or verifying reality.
Techniques like RLHF and prompt engineering can reduce some errors but cannot fully eliminate unpredictable hallucinations, so reliable use often requires careful prompting or external verification of answers.