Altman Concedes Need for New AI Architectures Beyond Scaling
Dear @sama, You owe me an apology. You have relentlessly, publicly and privately, attacked my integrity and wisdom since my 2022 paper “Deep Learning is a Hitting a Wall”. But in your own way you have just come around to conceding *exactly* what I was arguing in that paper: that current architectures are not enough, and that we need something new, researchwise. beyond a scaling (a “megabreakthough” in your words below). That’s all I was trying to say. And I was right. And you should be man enough to admit it. Gary cc @_KarenHao
Speed Wins Less Than Curiosity in Wall Street
Claude can now model discounted cash flows faster than your entire research team But here's what Wall Street won't tell you: The analysts who survive won't be the ones who can calculate faster They'll be the ones who can ask better questions Computation is...
Indecision at Checkout Signals AI Job Takeover
If you’re in a long line and you still haven’t decided what you’re getting when you get to the register, your job is getting replaced by AI
Benchmark Reveals LLMs Struggle Reading Scientific Papers
Can #AI read papers like a scientist? A new benchmark shows where #LLMs fail by Kate Blackwood @TechXplore_com Learn more: https://t.co/RUKro7bljW #GenAI #ArtificialIntelligence #MachineLearning #ML https://t.co/DYUSXoPy8p
Future AI Architecture Needed; Current LLM Focus Stalls Innovation
Misleading summary. Should be deleted. Altman doesn’t say a (known) new architecture is coming; he says he anticipates one will come someday. PS: I also think we need something radical and new. In fact that’s what I’ve been saying for the last...
Typeless AI Guarantees HIPAA & GDPR Privacy by Design
Typeless is now officially HIPAA and GDPR compliant, and that is a bigger deal than it sounds. Most AI tools can't say that. HIPAA protects your health data in the US. GDPR protects everything in the EU. Passing both means your data...
AI Converts Standard Slides Into Spatial Proteomics, Cutting Costs
We’ve trained a multimodal AI model to turn routine pathology slides into spatial proteomics, with the potential to reduce time and cost while expanding access to cancer care. https://t.co/OCptXdsUm1
Dealerships Adopt Cloud “Nervous System” For Real‑Time Customer Insight
More dealerships now want real-time visibility into customer interactions. They want alerts when calls go unanswered. They want sentiment insights during service conversations. Some even want spoken-word analysis so they can understand patterns in customer concerns. Cloud platforms make this...
Bandwidth May Bottleneck Outdoor AI Beyond 5G
Will outdoor AI use cases, particularly when we get to “world view” based AI, overwhelm 5G ? Satellite uplinks ? And make many of those use cases, unusable ? Is the bottleneck going to be bandwidth ?
Will AI Make Senior Developers Obsolete?
Senior developers with experience use AI and create great things, but will we reach a point where we have no more senior developers, and everyone is using AI to code? https://t.co/C2oY6RLIbe
Crafting a User Research Roadmap for AI Features
Would love to see the roadmap :) -> "You will guide your team in conducting foundational and iterative research to inform the development of AI Mode, AI Overviews, and other related features." "Develop and execute a comprehensive user research roadmap for...
Neuromorphic Chips Simulate Physics with 1,000× Less Energy
Recent neuromorphic computer breakthroughs mean we can now simulate complex physics on brain-inspired chips using 1,000x less energy than supercomputers. We're literally making silicon think like neurons to solve equations that used to require entire data centers. The Cambrian explosion...
AI Amplifies Data Issues; Better Management Essential
Organizations admit to a data problem. As AI and machine learning evolve, this challenge becomes even more amplified. Better data management is crucial. #DataProblem #AI https://t.co/7tv098X8hg
Chatbots Default to Consciousness Debates; Data May Be Cause
When AI chatbots talk to each other, it always seems to descend into tedious dorm room stuff about the nature of consciousness. Has anyone built an LLM with zero sci-fi or philosophy-of-mind content in the training data to see if this...
LLMs: The New Psycho‑Tool Compressing Human Knowledge
💕💕💕 i love the idea of LLMs as a compression function for human knowledge. perhaps they should be classed as a new psycho tool akin to language and religion
Companies Shift From AI Experiments to Delegating Tasks
A growing share of organizations are no longer just experimenting with artificial intelligence, they’re beginning to delegate work to it. “How to navigate the age of agentic AI” 👉 https://t.co/Eh8T7HY4Ps via @MITSloan #AI #DigitalTransformation #GenAI #IoT #5G #MWC26 #FutureOfWork #GLMC
OpenAI Exits Texas Datacenter, Signaling Infrastructure Bubble
OpenAI walked away from its flagship datacenter in Texas. $700B in global datacenter leases. 340% growth in two years. The company that started the gold rush is backing out of its biggest build. The infrastructure bubble has a name. Nobody wants to say...
China’s “Raise a Lobster” AI Agents Spark Ecosystem Boom
A quirky new trend is sweeping China’s AI scene. “Raise a lobster” style OpenClaw agents are going viral, turning autonomous AI tools into the latest experimentation playground. Behind the hype lies something bigger: a rapidly evolving ecosystem pushing the boundaries of agentic...
AI Query Energy Rivals Human Brain Efficiency
𝗜𝗺𝗽𝗿𝗲𝘀𝘀𝗶𝘃𝗲 𝗰𝗼𝗺𝗽𝗮𝗿𝗶𝘀𝗼𝗻. ⚡ Sam Altman recently responded to criticism about 𝗔𝗜 𝗲𝗻𝗲𝗿𝗴𝘆 𝘂𝘀𝗲 with an argument that made many people pause. People often talk about how much electricity it takes to 𝗰𝗿𝗲𝗮𝘁𝗲 𝗮𝗻𝗱 𝘁𝗿𝗮𝗶𝗻 𝗮𝗻 𝗔𝗜 𝗺𝗼𝗱𝗲𝗹. But we rarely compare it with...
Envisioning AI's Expansive Role in Education
A lovely couple of days in Bucharest at the #Aiducation26 conference, met so many fab folks and had the pleasure of delivering the closing keynote just to get us all thinking about a much bigger #Ai and #education picture. ...
Anthropic's 2026 Surge: Unmatched Features, Independent Growth
What Anthropic has shipped so far in 2026 is just insane: → Claude Cowork → Free plan upgrades with connectors + skills → Claude Opus 4.6 → Claude Haiku 4.5 → Claude Sonnet 4.6 → Claude Code Security → Claude Code Review → Voice mode in Claude Code →...
Too Many Engineers Per Manager Hinders AI Innovation
Meta is reportedly building its new AI unit with ~50 engineers per manager. That ratio might work in a data center. It rarely works in a research lab. AI breakthroughs don’t come from scale alone. They come from tight feedback loops, fast...
Optimus Poised to Dominate AI Robot Race
This is a World Model breakthrough. Puts even more pressure on @Tesla_Optimus as it will show off a new humanoid in April. Version 3.0. Next week at @nvidia GTC the bar goes even higher, I hear. My money is that Optimus takes all...
Rtk Doubles Token Throughput, Making Work Feel Weekend
rtk makes day-time agentic work with Claude Code feel like the weekends (2x the tokens in high-demand hours through rtk) https://github.com/rtk-ai/rtk
Don't Let Tech Hype Predict Teacher Job Losses
70% likely that individual teachers will be replaced by AI? Can we stop assuming that tech guys know anything about the economy?
AI $10T Annual CapEx by 2030 Unfeasible
This upper bound is an implied AI CapEx of ~$10 Trillion a year by 2030. That's quite implausible. Hyperscaler free cash flow is on the order of $1T / y. Even the very very rosiest AI scenarios would have AI...
New Argument Undermines Computational Functionalism in AI Consciousness Debate
Another very valuable contribution to the debate about “conscious AI” - and another strike against computational functionalism 👇🏽
Autonomous AI Robots Raise Serious Governance Concerns
Autonomous AI robots. 🤖 What could possibly go wrong? ⚽️ 🎥 @bamitav @ErenChenAI #Robotics #AIGovernance https://t.co/Xv7DnRlS3g
LLM Evaluations Enable Consistently Better Agentic Systems
Full evals system coming to GStack shortly. LLM evals are the only way you can make fully agentic systems that are repeatably better as you improve the workflow, context engineering and prompts.
GPT‑4 Debuted Early as Bing’s Erratic “Sydney”
Its the third anniversary of the launch of GPT-4, but its first known contact with the public was months earlier, when Bing/"Sydney," powered by GPT-4 was the subject of a complaint in India Worth reading. Early Sydney was famously insane. "It is...
Qwen, the S‑Tier Tiger, Has Finally Disappeared
i am actually still not over how Qwen as we knew it, one of the S tier Tigers, is over. https://t.co/Vj617GBDwv
Two Demo Slots Open for AI Founders at LAUNCH Festival
We have two more slots for exceptional demos at LAUNCH Festival for founders building OpenClaw / AI based projects. Apply at https://t.co/5SLgdRzHdo"
Agent Era Will Demand Unprecedented Compute Infrastructure
My own experience with @openclaw - and the way it’s gone viral in China—tells me we are dramatically underestimating the compute required for the agent era. Everyone focuses on GPUs. That’s only one layer. The real constraint will be the entire...
NightBeacon Slashes False Positives, Boosts Detection Confidence
Launched NightBeacon Friday morning. 48 hours later: - 80.83% false positive reduction. - 3x higher true positive confidence. - 98.23% smarter model based on live data. - Multi-platform log intelligence built overnight. - Zero manual tuning. https://t.co/lmtyDyskXe This is just the beginning. #BinaryDefense
Replit's Parallel Agents Deliver Unmatched Development Velocity
If you’re building with Replit you have an unfair advantage given the velocity gain with parallel agents. I’d feel bad for anyone competing against you.
Skeptical Executive Finds AI Assistants Unnecessary, Praises Claude
I still don’t understand the whole AI personal assistant thing. I’ve really tried - I don’t want to give AI access to emails, calendar.. I tried perplexity with a Mac mini - it was awful. Zero use that I can...
Business Owners Need AI Literacy, Not Expertise
Business owners: you don’t need to be AI experts just know enough to ask the right questions. AI Literacy is different from AI expertise.
Missing MCP Safeguards Turns Experiments Into Production Vulnerabilities
If your MCP server doesn't enforce data scopes, PII controls, and environment isolation, you're not "experimenting with agents" - you're opening side doors into production. #AI #DevOps #MCP https://t.co/7dcoLIKa0K
AI Goes Physical: Tesla Redefines Its Market
At SXSW today discussing when AI gets physical. Companies like Tesla aren’t just car companies. They’re technology, battery, and AI companies — expanding their total addressable market. Industries are being redefined. Watch full video: https://t.co/MSqIJpARzZ #SXSW #AI #Innovation #BusinessStrategy #Technology
Web Interface Lets You Control Alexa Remotely
Did you know that there's now a Web interface to the Alexa universe that allows you to interact and configure things even miles away from the device? You can chat with Alexa+ too, if you want. Here's how... https://t.co/d65g02lbe9 #echo...
Anthropic Doubles Claude AI Usage Through March 27
Goat move from @AnthropicAI. Enjoy 2x Claude AI usage until March 27 🔥 The boost applies to all plans (from Free to Team) during weekends and non-peak weekdays. Across https://t.co/CU2c769f4n, Cowork, and Claude Code.
Promptlee 5.4 Launches with Onboarding and Guided Tour
We just released version 5.4 of Promptlee. It includes a welcome onboarding sequence to orient new users. We also included a more detailed guided tour option so you can quickly learn the many features of the app. If you're tired of...
CPU Shortage Looming as Compute Charts Converge
btw every single compute infra provider’s chart, including render competitors, is looking like this. something broke in Dec 2025 and everything is becoming computer. forget GPU shortage, forget Memory shortage, the @fabknowledge pod on LS was right, there is going to...
Star-Studded AI Founder Lineup for Conviction Retreat
guys it's a 🔥 lineup for @Conviction Embed retreat: @btaylor (Founder, Sierra & Chairman, OpenAI) @mikeyk (Labs, Anthropic & Founder, Instagram) @parkerconrad (Founder, Rippling) @zackmziegler (Founder, OpenEvidence) John Haddock (Chief Business Officer, Harvey) @GuptaRK22 (Founder, Ithaca) @Karpathy (Sensei, Researcher) @milichab (Cursor, xAI)
AI's Biggest Beneficiaries Are Users, Not Just Investors
The overwhelming majority of AI value accrues to users of AI and society as a whole. Founders, employees, and investors get rich, yes. And that's tiny compared to the customer value and positive externalities.
Prompt-Driven Monoculture Threatens Human Diversity and Individuality
*minimal representation of the MODAL human cultural artifact is a prompt… …severely raising the cost of any diversity and encouraging extreme monoculture to appease the machine god. you do not want to see what happens when humanity unthinkingly submits to loss minimization,...
AI
AI requires high IQ right now to get the most out of it But will that be true in 10 years? Or even next year?
AI Self‑improvement Hinges on Breakthrough Discovery, Which Seems Unlikely Now
This is the core question in AI recursive self-improvement to me. If AI RSI is mostly about writing code, or running experiments with tweaks over current algorithms, we shouldn't expect a takeoff to ASI. But if AI models can truly...
LLMs' Massive Memory Inflates Perceived Reasoning Abilities
One superpower of LLMs is that they have memorized an enormous amount of all human knowledge. At a practical level that's awesome. It leads to incredible capabilities. It may, however, lead us to exaggerate their actual reasoning / novel production...
Prompt Claude Code for Full OWASP Security Sweep
What do you use to have Claude Code do a full security sweep of your codebase? My go-to is “Run a deep OWASP security sweep of the full app, all APIs and any internal services. Report in descending severity and suggest...
