@emollick Avatar @emollick Ethan Mollick

Ethan Mollick posts on X about ai, this is, claude code, llm the most. They currently have [-------] followers and [---] posts still getting attention that total [-------] engagements in the last [--] hours.

Engagements: [-------] #

Engagements Line Chart

Mentions: [---] #

Mentions Line Chart

Followers: [-------] #

Followers Line Chart

CreatorRank: [------] #

CreatorRank Line Chart

Social Influence

Social category influence technology brands 24.41% social networks 3.76% stocks 2.35% finance 2.35% countries 1.41% gaming 0.94% events 0.47%

Social topic influence ai #1479, this is 6.57%, claude code #66, llm #123, the first #3516, seedance #282, the new #3900, future #713, open ai #989, the most 3.76%

Top accounts mentioned or mentioned by @nabbilkhan @ghidorah_x @alygg77 @kongkou_ @mingtakaivo @lfuckingg @andythinkmode @udaysy @brandgrowthos @victor_uwer @wildpinesai @urbanecholot @stefan171 @4484 @grok @deredleritt3r @krishnanrohit @powerofaii @wakewcamon @huanghlzone

Top assets mentioned Alphabet Inc Class A (GOOGL)

Top Social Posts

Top posts by engagements in the last [--] hours

"I asked Opus [---] in Claude Code to build me Borges's Library of Babel. An hour later it delivered this. The AI implemented a Feistel cipher so each possible book has a real location you can search for text across the Library & find the book. Enter it: https://library-of-babel-3d.netlify.app/ https://library-of-babel-3d.netlify.app/"
X Link 2026-02-06T07:02Z 320.8K followers, 115.1K engagements

"People are more resilient to major negative life events than many think. 😐The graphs show that generally life satisfaction bounces back faster than expected after bad events. 💪This review gives an overview on the many pathways we have for resilience. https://www.public.asu.edu/iacmao/PGS191/Resilience%20Reading%20%231A.pdf https://www.public.asu.edu/iacmao/PGS191/Resilience%20Reading%20%231A.pdf"
X Link 2023-10-27T02:31Z 319K followers, 56.5K engagements

"Online research surveys may be a dying methodology as techniques for verifying respondents as humans are failing: Automatic AI detection systems are currently completely unusable Individual attention checks will no longer be a sufficient tool to ensure good data quality"
X Link 2024-10-21T13:58Z 320K followers, 57.3K engagements

"My impressions of getting to work with a preview of the new Claude with computer use and putting it to work across multiple tasks. You can see the promise of agents but also where they still need work. I think it is a sign of the near future. https://www.oneusefulthing.org/p/when-you-give-a-claude-a-mouse https://www.oneusefulthing.org/p/when-you-give-a-claude-a-mouse"
X Link 2024-10-22T18:20Z 320K followers, 32.4K engagements

"This feels pretty big to me. There is an assignment I give my MBA students where I give them a financial model for a business & ask them to identify missing or incorrect assumptions. The new Claude analysis module does a really good job spotting the issues & revising the model"
X Link 2024-10-25T16:44Z 320K followers, 46.8K engagements

"LLMs have been quickly climbing the scale of human ability passing tests reserved for graduate students exceeding people on narrow tasks Yet there is no coherent intellectual job where LLMs exceed the top humans. I see many people assume this will change for which jobs When"
X Link 2024-10-28T01:01Z 320K followers, 78.2K engagements

"GPT-5.2 Pro is good enough to check reproducibility & robustness of academic papers across many fields (given the data can you get the same results are the statistics brittle). At scale this would have a big impact. It can't do an independent replication with new data yet"
X Link 2026-01-22T22:01Z 319.4K followers, 131.6K engagements

"The difference between the software being the point (Microsoft's perspective) and the output being the point (Anthropic's perspective). Do you care about the tool or about the answer Seems like that is going be a very consequential decision for who wins the agentic work space"
X Link 2026-01-24T02:45Z 320.6K followers, 24K engagements

"Canaries in the coal mine. Worth paying attention to. (And yes they are both obviously interested in seeing their own products used but hearing enough from other independent coders that make me believe them. I wrote more about the shift here: https://www.oneusefulthing.org/p/management-as-ai-superpower https://www.oneusefulthing.org/p/management-as-ai-superpower"
X Link 2026-01-28T20:54Z 318.4K followers, 109.4K engagements

"Generally I find people who are anti-AI because they think AI is just hype respond rather well to being shown how AI works in a non-judgmental way. And if they hate AI for other moral reasons they may still dislike AI after but they become more informed critics & we need those. @iammattduff do you think theyd be less anti AI if they knew about the true capabilities or just madder @iammattduff do you think theyd be less anti AI if they knew about the true capabilities or just madder"
X Link 2026-01-29T00:10Z 318.4K followers, 35.5K engagements

"A good sociological explanation of why that odd "95% of AI projects fail MIT study" (that was not actually a study at all but based on someone's unexplained interpretation of [--] unspecified interviews at a conference) somehow became a ubiquitous point of discussion last summer"
X Link 2026-01-29T02:17Z 319.8K followers, 37.9K engagements

"I warned you all to pick a better name but noooo. Now everyone is going to be talking about agent swarms at conferences and you are going to have a corporate swarm leader running swarm initiatives and swarm training. Flocks are cute we should have called them flocks. Unfortunately the moniker swarm is here cc @emollick https://t.co/TOkIvthbeS Unfortunately the moniker swarm is here cc @emollick https://t.co/TOkIvthbeS"
X Link 2026-01-29T07:18Z 318.9K followers, 51.5K engagements

"Had early access to Genie [--] world modelling. Huge leap forward in modelling/physics but some issues remain Here is a bit of an otter airline pilot with a duck on its head walking through a Rothko inspired airport and an otter in a wingsuit flying through a city of gothic towers"
X Link 2026-01-29T17:02Z 318.7K followers, 97.5K engagements

"Can Genie [--] run Doom "Doom but the walls are made entirely out of screens also running Doom featuring the marine from Doom but also his head is a screen running Doom""
X Link 2026-01-29T21:10Z 319.5K followers, 458.3K engagements

"OK this amused me. I gave Claude a copy of this classic XKCD comic and the prompt: "solve this. do a great job. ask no questions." And it gave me this in one shot [--] minutes later including the style (and the note that it thought it was 2024): https://xkcd-1425-bird.netlify.app/ https://xkcd-1425-bird.netlify.app/"
X Link 2026-01-30T03:06Z 319.1K followers, 179.6K engagements

"Its very hard to know what is real with 🦞bot right now lots of stuff that may be emergent behavior or storytelling or prompt injection or who knows As impressive as these examples are the models are still Claude [---] Opus which is great but also has obvious limits to ability guys. its happening. Henry - Alex's clawdbot(moltbot) - worked overnight while alex was asleep and: acquired a phone number took the elevenlabs api and set u a voice connected to his new phone number waited for Alex to wake up and started calling him. can control guys. its happening. Henry - Alex's clawdbot(moltbot) -"
X Link 2026-01-30T20:35Z 318.5K followers, 59.4K engagements

"Been pasting Calvino's Invisible Cities verbatim into Genie and it works surprisingly well: Marco Polo in Octavia the web city over a chasm; Armilla the city of pipes; Ersilia the city where threads between buildings show relationships; & Sophronia half carnival half "real""
X Link 2026-01-31T06:07Z 318.4K followers, 47.1K engagements

"Stochastic parrot was an amazing turn of phrase that was (& still somewhat is) technically correct without being illuminating about current LLMs It shows both the power of analogies and highlights the failure to come up with something nearly as good that explains LLM capability"
X Link 2026-01-31T19:25Z 319K followers, 29.4K engagements

"Fun to turn paintings into scenes I can walk around in using Genie 3: here are the works of Giorgio de Chirico Munch Turner and the Bayeux Tapestry. I can move freely around the scenes and yes they are a little weird but its real-time dynamic image creation by the AI"
X Link 2026-02-01T06:46Z 319.1K followers, 56.6K engagements

"Moltbook seems to have "broken through" to a wider non-AI audience A good chance to explain the real present (this time Moltbook was mostly roleplaying by people & agents) & the risks for the future (independent AI agents coordinating in weird ways spiral out of control fast)"
X Link 2026-02-01T19:08Z 319.5K followers, 28.3K engagements

"The many eulogies for AI capability growth after the release of GPT-5 seem especially short-sighted right now. Letting people nervous about AI feel they can safely ignore AI development because it was pure hype that would never have any real impact is not a good thing for anyone"
X Link 2026-02-02T01:56Z 319.5K followers, 27.2K engagements

"This is a good thread and a useful one. .but you should not assume that the Claude Code team has discovered the actual best ways to use Claude Code. A lesson of the past three years is that figuring out the best way to use AI is an exploratory process that is open to many I'm Boris and I created Claude Code. I wanted to quickly share a few tips for using Claude Code sourced directly from the Claude Code team. The way the team uses Claude is different than how I use it. Remember: there is no one right way to use Claude Code -- everyones' setup is I'm Boris and I created Claude Code. I wanted"
X Link 2026-02-02T03:07Z 318.7K followers, 48.1K engagements

"Took an old photo of a WWI battlecruiser gave it to Genie [--] and prompted it to let me play as a torpedo boat at the Battle of Jutland. Considering this is a research preview astonishing how fast this has come. An AI dynamically generating the world with no game engine"
X Link 2026-02-02T05:43Z 318.5K followers, 30.4K engagements

"On one hand it is reasonable to take announcements about what AI can do from AI companies with a grain of salt. On the other this fits the trajectory we are seeing in AI coding everywhere. I think the implications are pretty large for software development and beyond. Codex now pretty much builds itself with the help and supervision of a great team. The bottleneck has shifted to being how fast we can help and supervise the outcome. Codex now pretty much builds itself with the help and supervision of a great team. The bottleneck has shifted to being how fast we can help and supervise the outcome"
X Link 2026-02-02T14:05Z 319.5K followers, 27.8K engagements

"I still see the argument that AI is bad af generating ideas but my colleagues and I along with many other researchers have repeatedly found that even the old GPT-4 could be prompted to generate more diverse and higher quality ideas than most people. And newer models do better. 🚨 Our new paper: we know that GPT-4 generates better ideas than most people but the ideas are kind of similar & variance matters But it turns out that better prompting can generate pools of good ideas that are almost as diverse as from a group of humans https://t.co/LkGsU0VC7S https://t.co/5BDx8A3E8Z 🚨 Our new paper:"
X Link 2026-02-02T18:19Z 319.7K followers, 32.7K engagements

"Anthropic has been releasing an impressive array of papers recently using a variety of methods most of which show potential AI issues rather than just cheerleading about AI. Also they tend to be very well communicated (with a whiff of Claude about the writing to be sure). Finding 1: The longer models reason the more incoherent they become. This holds across every task and model we testedwhether we measure reasoning tokens agent actions or optimizer steps. https://t.co/3VkfVESNiM Finding 1: The longer models reason the more incoherent they become. This holds across every task and model we"
X Link 2026-02-03T01:27Z 318.8K followers, 139.5K engagements

"Yes OpenAI has more consumer users but they have well over a million paying enterprise users and a lot more API use so I think this is not just a Anthropic is enterprise/ChatGPT consumer thing"
X Link 2026-02-03T02:08Z 318.3K followers, [----] engagements

"It is worth worrying about the formation of a single omincausal "anti-AI" coalition that can only agree on a complete halt to AI as a remedy. Not only is that unlikely but it undermines the desire to make policies that channel AI to good uses or that mitigate specific harms. No single AI complaint/fear is salient enough to enough people to form a durable political movement so what is happening instead is that an omnicausal anti-AI sentiment is forming. Kids and electricity and water and jobs and dontkilleveryoneist memes and also it hallucinates. No single AI complaint/fear is salient enough"
X Link 2026-02-03T17:11Z 319.7K followers, 18.6K engagements

"Increasingly the big labs need their apps to come with good models and good harnesses. The most common app is the chatbot. The Big [--] all have good models and good chatbot apps but Gemini lacks a good harness. IDEs/CLIs are the new battleground Knowledge work apps are next. Gemini not being in the conversation at all with Claude Code and Codex is the real code red emergency. Gemini not being in the conversation at all with Claude Code and Codex is the real code red emergency"
X Link 2026-02-03T19:15Z 318.3K followers, 25.8K engagements

"I am not sure what was better: All the roleplaying slop replies coming from badly prompted LLMs pretending to be people or the fact that I am now seeing roleplaying slop replies from self-professed 🦞bots. Its crustaceans all the way down either way"
X Link 2026-02-03T19:34Z 318.4K followers, 15.1K engagements

"Hard to know which X articles are valuable but this is a good summary of the significance of world modeling by a distinguished scientist and robot expert NVIDIA https://t.co/Npar79SvUh https://t.co/Npar79SvUh"
X Link 2026-02-03T19:53Z 318.8K followers, 72.7K engagements

"I dont want my sycophantic Clawbot calling me for reassurance but the interesting thing here is that the tweet is the instructions for the agent to set itself up. Plain English instructions that agents can follow may be a new avenue for marketing (and a security nightmare) https://t.co/AYYFtmBZUO https://t.co/AYYFtmBZUO"
X Link 2026-02-03T23:20Z 319.6K followers, 62.1K engagements

"GPT-5.2 Pro and Thinking spot the issue in Anthropics analysis skill Opus [---] and Gemini [--] Pro do not (Its a not huge issue but interesting) (The [---] models have the strongest statical reasoning in my experience & [---] Pro is best if you want pedantic correct second opinions) Confidence intervals are so confusing that even Anthropic's skills documents get them wrong https://t.co/A0ewFfmk8v Confidence intervals are so confusing that even Anthropic's skills documents get them wrong https://t.co/A0ewFfmk8v"
X Link 2026-02-04T15:42Z 318.9K followers, 71.4K engagements

"Finally some data on Grok is this true Grok is reasonably accurate and meaningfully shifts opinions but its use is polarized politically Signs that even our AI preferences may be subject to increasing ideological tribalism when it comes to news and information. 🚨New WP "@Grok is this true" We analyze 1.6M factcheck requests on X (@grok & @Perplexity): 📌Usage is polarized Grok users more likely to be Reps 📌BUT Rep posts rated as false more ofteneven by Grok 📌Bot agreement with fact-checks is OK but not great; APIs match https://t.co/UjgDXgohib 🚨New WP "@Grok is this true" We analyze 1.6M"
X Link 2026-02-04T21:02Z 318.4K followers, 20K engagements

"As jobs begin to shift with AI I think there will be increasing numbers of people feeling like Aditya. I think figuring out what comes next is challenging. People are more psychologically resistant to major changes than you might expect but that doesnt mean it will be easy. People are more resilient to major negative life events than many think. 😐The graphs show that generally life satisfaction bounces back faster than expected after bad events. 💪This review (https://t.co/lGLPIaaRPI) gives an overview on the many pathways we have for resilience. https://t.co/2diMDuIzZx People are more"
X Link 2026-02-04T21:16Z 318.4K followers, 27.8K engagements

"You know you are living through a discontinuity in a field when almost all of the science fiction about AI is obviously off (with a few notable exceptions) Like reading early 1900s science fiction where everything is about the canals of Mars right after Mariner actually landed"
X Link 2026-02-05T01:10Z 320.5K followers, 31K engagements

"@deredleritt3r Lena Accelerando (sort of at least for the spirit of living through the thing) for the far future of living with smart minds Bank's Culture series obviously. https://qntm.org/mmacevedo https://qntm.org/mmacevedo"
X Link 2026-02-05T01:14Z 320.5K followers, [----] engagements

"@deredleritt3r Also Blindsight. Not in good ways"
X Link 2026-02-05T02:54Z 320.5K followers, [---] engagements

"Chinese open weights models remain the same 7-9 months behind but o3 level is an achievement I still don't get the business model of open weights companies as costs go up however. You can't make money from services or selling ancillary products like traditional open source Kimi K2.5 set a new record among open-weight models on the Epoch Capabilities Index (ECI) which combines multiple benchmarks onto a single scale. Its score of [---] is about on par with o3 Grok [--] and Sonnet [---]. It still lags the overall frontier. https://t.co/tKNPKWIo3Y Kimi K2.5 set a new record among open-weight models on"
X Link 2026-02-05T03:31Z 319K followers, 47.1K engagements

"We need a moratorium on the following depictions of AI: gleaming white robots floating blue holographic brains glowing amber lines connecting a dark globe Lawnmower Man style 1990s computer graphics semitransparent screens floating in the air with vector art faces on them"
X Link 2026-02-05T11:13Z 320.5K followers, 31.5K engagements

"Had early access to Opus [---] here is the results from same prompt to create the control panel for a spaceship in the distant future done using the standard Claude interface (not Code). I had to zoom into a few of the subsystems so you can see the details. Had early access to Gemini with Deep Think. Very good model big gains over standard Gemini [---] Pro for a lot of problems. Here is the first attempt at the starship control panel prompt I try with every model. First time I have seen a model make a 3D interface in response. https://t.co/bLFF2IcOP3 Had early access to Gemini with Deep Think."
X Link 2026-02-05T18:09Z 320.5K followers, 106.6K engagements

"Opus [---] saturates my Lem test which I've done since GPT-3.5 SciFi author Stanislaw Lem wrote of two rival constructors of robots. One creates a robotic poet & the other challenges it to write an impossible poem it does Opus does it as a [--] line poem sonnet & 🤯 a sestina. Grok [--] passes the Lem test first try with the most coherent narrative yet. https://t.co/Pn92JL1JBC Grok [--] passes the Lem test first try with the most coherent narrative yet. https://t.co/Pn92JL1JBC"
X Link 2026-02-05T18:28Z 320.5K followers, 53.5K engagements

"We are going to need more compute now that agents can complete long-term economically valuable tasks. (Derek has done some of the most careful takes on is AI a bubble) for me the odds that AI is a bubble declined significantly in the last [--] weeks and the odds that were actually quite under-built for the necessary levels of inference/usage went significantly up in that period basically I think AI is going to become the home screen of a for me the odds that AI is a bubble declined significantly in the last [--] weeks and the odds that were actually quite under-built for the necessary levels of"
X Link 2026-02-05T19:20Z 320.5K followers, 54.5K engagements

"Data confirms that Opus [---] was a super weird model"
X Link 2026-02-06T00:49Z 320.2K followers, 15.3K engagements

"Its interesting that theres a whole bunch of people trying to figure out how you do SEO for AI models that do not like being manipulated that interact in ways we dont fully understand with context and with user preferences & know when theyre being measured. Seems hard. The Opus [---] system card has some extremely wild stuff that remind you about how weird a technology this is. These paragraphs are really worth reading. https://t.co/Ybpx8Egjxm The Opus [---] system card has some extremely wild stuff that remind you about how weird a technology this is. These paragraphs are really worth reading."
X Link 2026-02-06T02:28Z 320.1K followers, 23.3K engagements

"With the current state of AI there is a lot of alpha as they say in being a good manager who understands process can explain it clearly knows what info is needed for a task and has expertise in evaluating the results of work quickly. These are transferable skills to AI"
X Link 2026-02-06T03:23Z 320K followers, 22K engagements

"Feel free to use this in your next presentation"
X Link 2026-02-06T05:36Z 320.1K followers, [----] engagements

"Original prompt: "build me Borges's Library of Babel. Make it as impressive and accurate as you can. Decide how you want to handle things dont ask questions but test and improve it until it will blow me away. You can use the image creation agent if you need it. this should be an original implementation not an extension of another work" I also asked it to add the vestibules. https://twitter.com/i/web/status/2019668757841367491 https://twitter.com/i/web/status/2019668757841367491"
X Link 2026-02-06T07:05Z 319.5K followers, 16.9K engagements

"Within the library that Opus implemented are books that contain all of the weights for Opus [---] the cure for cancer and a perfect prophecy of the future. (And orders of magnitude more things that look like the weights the cure and a true prophecy but are actually wrong.)"
X Link 2026-02-06T14:27Z 320.1K followers, 13.8K engagements

"The Opus weights will be spread across many volumes of course. But in one of the books is likely a single sentence that will change your life. You just need to find it. Code: https://github.com/emollick/library-of-babel https://github.com/emollick/library-of-babel"
X Link 2026-02-06T17:19Z 320K followers, [----] engagements

"@JosephByDesign Naw been doing library of babel stuff for awhile. I couldnt get it to work in Gemini [--]. Good taste though"
X Link 2026-02-06T18:28Z 319K followers, [----] engagements

"What is the deal with Futurism They seem to deliberately distort information about AI news. Criticism of AI is important but their angle is "AI isn't real" and the articles are regularly shared to make that point Here is an example of a headline they ran (and quietly updated)"
X Link 2026-02-06T21:22Z 320K followers, 16.3K engagements

"This paper (by a colleague) is being misunderstood - it is not a prediction about whether the METR exponential curve is ending but an argument that it is plausible to model the exponential as a series of stacked s-curves which would require further breakthroughs (at the scale of reasoning) to continue. I don't necessarily even think this contradicts what most insiders in AI from Demis to Andrej are saying I wanted to make a few clarifications which we believe were clear in our paper but not in my original post (re-analyzing @METR_Evals data). Our contribution is to posit progress as a"
X Link 2026-02-06T21:34Z 320.4K followers, 26.2K engagements

"One conversation centered on the ever accelerating progress of technology and changes in the mode of human life which gives the appearance of approaching some essential singularity in the history of the race beyond which human affairs as we know them could not continue. [----] Obituary for John von Neumann by Stanislaw Ulam https://twitter.com/i/web/status/2019913568204763453 https://twitter.com/i/web/status/2019913568204763453"
X Link 2026-02-06T23:18Z 320.1K followers, 15.8K engagements

"Mac only releases for AI tools is an annoying trend. This is an opinion that unites both gamers and corporate users"
X Link 2026-02-07T13:12Z 320.6K followers, 32.8K engagements

"Yes. I am looking at you Cowork and Codex"
X Link 2026-02-07T13:15Z 320.5K followers, 13.6K engagements

"I wrote back in [----] that AI is creating four singularities for research (where a singularity is a point in human affairs where AI has so altered a field that we cannot imagine what the world on the other side of that singularity looks like). True now. https://www.oneusefulthing.org/p/four-singularities-for-research If youre an economist or any sort of social scientist Im not sure how you can look at the chart below and keep doing the same type of research you were a year ago. https://www.oneusefulthing.org/p/four-singularities-for-research If youre an economist or any sort of social"
X Link 2026-02-08T15:16Z 320.4K followers, 60.3K engagements

"The AI commentators have won on this platform not just because of the volume but because they produce meaning-shaped comments that take too long to parse and identify as meaningless. Posts are still useful commentary has become useless finding sparks of good stuff too hard"
X Link 2026-02-08T17:16Z 320.5K followers, 14.1K engagements

"@bygregorr The purchase is through Stripe so you don't give me any personal data though"
X Link 2026-02-08T20:39Z 319.2K followers, [---] engagements

"So much work is going into faking continual learning and memory for AIs and it works better than expected in practice so much so that it makes me think that if continual learning is actually achieved the results are going to really shift the AI ability frontier very quickly"
X Link 2026-02-09T04:16Z 320.5K followers, 50.4K engagements

"@krishnanrohit 100% Claude of course"
X Link 2026-02-09T04:17Z 319.1K followers, [---] engagements

"A corporate position that workers should "just use AI to do stuff" has never been enough. AI use in companies is a leadership problem that involves answering fundamental questions about what people should do with their time how work is organized and how to center people in work Interesting research in HBR today about how the productivity boost you can get from AI tools can lead to burnout or general metal exhaustion something I've noticed in my own work https://t.co/e0qocFYjL5 Interesting research in HBR today about how the productivity boost you can get from AI tools can lead to burnout or"
X Link 2026-02-09T17:12Z 320.5K followers, 45.4K engagements

"As an academic I am sympathetic as publishing takes awhile and it is hard to keep up with frontier models but. .especially if your argument is "AI is bad at X" you need to explain why you think it won't change graph any trend as models improve & update before publication i am begging academics to study AI capabilities using frontier models. the models used in this study (which is going to be cited for years as proof that "AI is bad at health advice") are GPT-4o Llama [--] and Command R+ two obsolete models and one i've never heard of. https://t.co/ekq502dhPe i am begging academics to study AI"
X Link 2026-02-09T18:09Z 320.5K followers, 33K engagements

"I don't think "judgement" is going to prove to be the bright line between human and AI work that people seem to expect it to be"
X Link 2026-02-09T21:38Z 320.6K followers, 37.1K engagements

"One big reason is that it is hard to verify really good writing so you can't do good RL on it. However I suspect that the above tweet will be quote tweeted sometime this year with an example of a very good piece of writing created by an LLM from scratch so we will see"
X Link 2026-02-10T00:40Z 320.2K followers, 11.7K engagements

"The paper actually has two big real points however: (1) Humans were bad at prompting (obsolete) AI to get medical advice - I suspect this is no longer as true (2) Benchmarks of medical knowledge don't always mean reality in serving patients. [--] has changed I think [--] has not"
X Link 2026-02-10T00:57Z 320.2K followers, [----] engagements

"You need one of GPT-5.2 Extended Thinking Opus [---] Thinking Gemini [--] Pro - yes you need to pay $20 Also I hate to say it because it is very expensive but GPT-5.2 Pro is the best for very hard high stakes problems"
X Link 2026-02-10T02:55Z 320.1K followers, 18.8K engagements

"Always found it interesting that the human eye can see colors that cannot be displayed on any screen or page. I had ChatGPT whip up a pretty good imaginary color viewer after asking it to review the scientific literature and getting the shades right. https://chatgpt.com/canvas/shared/698aa1744eb48191b7d95f3f744cacf7 For more color weirdness you can see imaginary colors by using the optical illusion template in this tweet: 🔵Stygian blue: a blue both totally blue & impossibly black 🔴Self-luminous red: a red that is brighter than white 🟠Hyperbolic orange: more orange than 100% orange color"
X Link 2026-02-10T03:10Z 320.1K followers, 32.9K engagements

"I think people on this site don't realize how much people's interactions with "AI" turn out to be when you ask them: customer service lines (which are almost certainly not GenAI yet) or Siri or maybe a free model (often via an off-brand "ChatPT" AI app they downloaded somewhere)"
X Link 2026-02-10T03:39Z 320.5K followers, 18.5K engagements

"The volume of the kinds of releases that actually impact real work has been accelerating. In the past couple days that includes OpenAI Frontier & new Deep Research Claude for Powerpoint and Cowork for Windows and the wider release of the solid Claude-powered MS agent for Excel"
X Link 2026-02-11T02:03Z 320.2K followers, 16.8K engagements

"Second video of the same prompt (Again this was two of two no retries)"
X Link 2026-02-11T03:17Z 320.5K followers, 12.9K engagements

"@MetacogniShane Better in every regard. But it isn't inconceivable that Google has stuff ready to go that could beat it"
X Link 2026-02-11T03:28Z 319.8K followers, [---] engagements

"This viral essay is worth reading. I agree that AI is a very big deal & that most people don't know how good it has gotten fast. A missing caveat is that AI is still quite jagged especially when it comes to work across teams & organizations which creates bottlenecks for now https://t.co/ivXRKXJvQg https://t.co/ivXRKXJvQg"
X Link 2026-02-11T16:49Z 320K followers, 124.6K engagements

"I agree most people far undervalue where AI is but I think the jaggedness lies not just in jobs/tasks but between them. There is a lot that happens when people start to work together or interface with the real world where AI fails either weirdly (timing sequenced events) or where it doesn't apply (client meetings). Changing those things may take a lot more time. https://twitter.com/i/web/status/2021629374949163518 https://twitter.com/i/web/status/2021629374949163518"
X Link 2026-02-11T16:56Z 320K followers, [----] engagements

"RT @emollick: @krishnanrohit This organization could have been a series of markdown files"
X Link 2026-02-11T19:06Z 320.3K followers, [--] engagements

"@J_K_Chesterton nice - link"
X Link 2026-02-11T20:31Z 320.1K followers, [---] engagements

"RT @paulnovosad: I wrote a decent paper with AI. It took me about [--] hours from start to finish including an interactive choose-your-own-bo"
X Link 2026-02-11T22:15Z 320.3K followers, [--] engagements

"@powerofaii I appreciate your advice I just can't actually figure out how to make actual subagents work documentation is sparse"
X Link 2026-02-12T03:55Z 320.1K followers, [---] engagements

"@AndyMasley Memory as implemented is a destroyer of value in using LLMs"
X Link 2026-02-12T06:21Z 320.5K followers, 17.6K engagements

"One more Seedance version of "Monica's apartment from the show Friends except all of the friends are otters wearing wigs. The otter with a Rachel wig says "Is anything weird" and the one with a Joey wig says "Nope all is normal"." Attempt [--] of [--]. The laugh track is something"
X Link 2026-02-12T02:39Z 320.8K followers, 118.8K engagements

"There are few very large organizations that have changed how they do business in any significant way due to GenAI yet. Organizational transformation is slow & hard But many organizations now have successful internal GenAI efforts in pockets of their companies which are scaling"
X Link 2026-02-12T15:39Z 320.8K followers, 14K engagements

"Every so often I go on a wild blocking binge and block everyone who puts out long and coherent-seeming replies within the first minute of a post. Then I remember it isn't worth doing"
X Link 2026-02-14T00:45Z 320.8K followers, 10K engagements

"2028 being vividly the year of "a nation of geniuses in a datacenter""
X Link 2026-02-16T04:47Z 320.8K followers, [----] engagements

"GPT generated this academic paper from a dataset in [--] minutes. It may not be a blockbuster but it tests something kind of interesting in a sophisticated way - in short it may be publishable somewhere. The text is prone to hallucinations but the data analysis less so. 1/2"
X Link 2023-05-04T02:10Z 320.8K followers, 209.1K engagements

"Four AI lab vague posting strategies: 1) vague emoji: 🙂🔜🦾 2) vague corporate: Big release at DevCon 3) vague riddles: the STARS are aligning this WINTER 3) vague Lovercraftian prophecy: I have seen the face of infinite eternities & it knows us: ph'nglui mglw'nafh technotopia Grandiose vagueposting on Twitter is the one tried and true marketing strategy for AI labs. But as it gets overused it eventually creates fatigue Grandiose vagueposting on Twitter is the one tried and true marketing strategy for AI labs. But as it gets overused it eventually creates fatigue"
X Link 2025-12-30T07:46Z 320.7K followers, 23.3K engagements

"Dynamically-generated world of Hamlet as seen from the perspective of the poison. Genie [--] is a lot of fun"
X Link 2026-01-29T17:50Z 320.7K followers, 130.1K engagements

"I get asked about Ralph Wiggum in business meetings. You didn't have to do this"
X Link 2026-01-29T19:05Z 320.7K followers, [----] engagements

"Playing as Godot finally arriving. Just as Beckett intended thanks to AI"
X Link 2026-02-02T06:41Z 320.7K followers, 65.1K engagements

"It is possible to get Genie [--] to make worlds that are visually interesting. I have been using images I generated in Midjourney of vast megastructures & odd cities in various styles. After [--] seconds I can freely wander around them. (Yes I controlled the cat in the first scene)"
X Link 2026-02-05T04:09Z 320.6K followers, 56.8K engagements

"Folks I can only test so many models a day even when given early access. If everyone would just slow down their release schedules for a half a decade or so the world can catch up"
X Link 2026-02-05T18:49Z 320.7K followers, 25.1K engagements

"(This does not mean that there couldnt be some sort of financial issue with financing the compute but does point to the idea that compute is not being overbuilt)"
X Link 2026-02-05T19:51Z 320.7K followers, 14.2K engagements

"A genuinely radical approach to software development with AI without any human intervention. Even if this approach doesnt work for many cases I think we need more leapfrogging visions for how to redo processes with AI: See also: https://www.danshapiro.com/blog/2026/01/the-five-levels-from-spicy-autocomplete-to-the-software-factory/ https://factory.strongdm.ai/ https://www.danshapiro.com/blog/2026/01/the-five-levels-from-spicy-autocomplete-to-the-software-factory/ https://factory.strongdm.ai/"
X Link 2026-02-07T13:37Z 320.7K followers, 37.4K engagements

"Why do so many X articles read like AI-written conspiracy theories/health hacks/LinkedIn productivity tricks that would fall apart with any serious research On second thought this isnt really a question"
X Link 2026-02-07T22:57Z 320.6K followers, 17.4K engagements

"The Claude [---] Opus UX automatically decides how long to think & it seems to suffer from the same issue as the early GPT-5 router not taking hard requests seriously when they are not specifically math/coding related The OpenAI system is complex but fine-grained controls helps"
X Link 2026-02-08T16:07Z 320.7K followers, 34.6K engagements

"I had Claude Code do this. It made [--] volumes that contain the weights for GTP-1 & a guide for doing inference by hand. I had it launch an online store that I think works with a limited run of [--] of Volume [--] sold at cost in the US (but at your own risk) https://weights-press.netlify.app/ Product idea for OpenAI (I know a lot of you follow me): an entirely paper-based LLM. Just [---] volumes and only [--] person years to do the math for the first token using the paper version of GPT-1 Give the weights actual weight. Plus an excellent setup for science fiction stories https://t.co/iDGetnej4H"
X Link 2026-02-08T20:11Z 320.8K followers, 96.4K engagements

"Sold out But I had Claude create and deploy all [--] volumes of The Weights to the site as well-formatted PDFs so you can download them for free if you want. [-----] pages in total. [---] million floating point numbers. This is everything that makes GPT-1. https://weights-press.netlify.app/ I had Claude Code do this. It made [--] volumes that contain the weights for GTP-1 & a guide for doing inference by hand. I had it launch an online store that I think works with a limited run of [--] of Volume [--] sold at cost in the US (but at your own risk) https://t.co/kTfH4MkSZL https://weights-press.netlify.app/"
X Link 2026-02-09T04:00Z 320.8K followers, 62.5K engagements

"The jagged frontier: "It messed up Rachel otter and Phoebe otter.""
X Link 2026-02-11T05:45Z 320.7K followers, [----] engagements

""In sum through an extensive (and costly) validation process we have demonstrated that GPT-5 mini performs very well at recovering the ground truth data. It is clearly better than highly trained graduate students at this specific information retrieval task." At 1000x less cost We coded our 100k articles using LLMs. Should you believe them To answer this we benchmarked [--] human RAs against [--] LLMs on their ability to recover ground truth article data. Details in the paper and appendices but the LLMs did well and handily beat the highly trained humans. https://t.co/64jU1CZJ9R We coded our 100k"
X Link 2026-02-11T20:54Z 320.7K followers, 33.8K engagements

"@hamandcheese generally do not plan beyond 5-10 years - the world will look so different that the planning will likely be irrelevant I think you should still save for retirement"
X Link 2026-02-14T00:30Z 320.6K followers, [---] engagements

"@GordonBrianR I would push back here. The models that are doing physics and math are generalized Reasoner LLMs. Scaling them alone has been largely sufficient to get us here. And I think agentic harnesses alone would be enough to do science using current models"
X Link 2026-02-14T19:31Z 320.6K followers, [--] engagements

"A pretty bold commentary in Nature written by linguists computer scientists and philosophers declaring "by reasonable standards including Turings own we have artificial systems that are generally intelligent. The long-standing problem of creating AGI has been solved.""
X Link 2026-02-03T03:17Z 320.8K followers, 155.3K engagements

"New paper from Google on accelerating science using Gemini (as opposed to its specialized research AIs). Lots of case studies and interesting advice on ways to work with these system. "We view the AI as a tireless knowledgeable and creative bright junior collaborator.""
X Link 2026-02-05T00:54Z 320.8K followers, 28.9K engagements

"I think agentic AI would work much better if people took lessons from organizational theory which has actually spent a lot of time understanding how to deal with complex hierarchies information limits and spans of control. Right now most agentic AI systems seem to pretend that models have basically unlimited ability to manage subagents when that is clearly not true. We need measures of spans of control for AI. A human tops out at less than [--] direct reports. I am pretty sure that [---] subagents is too much for an orchestrator agent - suspect we need middle management agents (yes I get it"
X Link 2026-02-08T01:06Z 320.8K followers, 146.9K engagements

"Of all of the dangers of AI papers this is most worrying: AI researchers building a tool to find new drugs to save lives realized it could do the opposite generating new chemical warfare agents. Within [--] hours it invented deadly VX and worse things https://www.nature.com/articles/s42256-022-00465-9 https://www.nature.com/articles/s42256-022-00465-9"
X Link 2022-07-19T11:22Z 320.8K followers, 14.4K engagements

"How email ruins your life: the expectation that you will always check your email outside of work hours is linked with negative health effects relationship issues & anxiety. Formal policies dont limit email stress instead it comes from the expectations of your bosses & peers"
X Link 2022-11-14T16:14Z 320.8K followers, [----] engagements

"Look if everyone is worried about students cheating on essays for AI instructors can just cheat right back. I asked OpenAI to give me an essay question & make a rubric for grading. I had GPT-3 actually write the essay. I then had the OpenAI grade the essay & give comments. ✅"
X Link 2022-12-02T18:25Z 320.8K followers, 15.1K engagements

"Tom Lehrer has put all his songs online (including lyrics & sheet music) and given away all rights to them. The site will only be up for a limited time. This is a very niche tweet that will make a small segment of people very happy. https://tomlehrersongs.com/ https://tomlehrersongs.com/"
X Link 2022-12-17T16:16Z 320.8K followers, 2.1M engagements

"Extraordinary new paper from Google on medicine & AI: When Google tuned a AI chatbot to answer common medical questions doctors judged 92.6% of its answers right compared to 92.9% of answers given by other doctors. And look at the pace of improvement https://arxiv.org/pdf/2212.13138.pdf https://arxiv.org/pdf/2212.13138.pdf"
X Link 2023-01-03T13:07Z 320.8K followers, 2M engagements

"I think we haven't fully absorbed the fact that careful academic papers have found ChatGPT clearly passes some of the most challenging American professional exams: 🩺United States Medical Licensing Exam 🎓MBA-level Operations exam 🧑⚖The Bar Exam (based on typical exam questions)"
X Link 2023-01-22T21:17Z 320.8K followers, 4.4M engagements

"🤯🤯Well this is something else. GPT-4 passes basically every exam. And doesn't just pass. The Bar Exam: 90% LSAT: 88% GRE Quantitative: 80% Verbal: 99% Every AP the SAT"
X Link 2023-03-14T17:51Z 320.8K followers, 10.7M engagements

"The new BloombergGPT AI may be harbinger of the next wave of corporate AI. Current AIs are trained on web data (though firms can add their own training) BloombergGPT is 52% either proprietary data or cleaned financial data. And it shows signs of being better at financial tasks"
X Link 2023-04-01T01:22Z 320.8K followers, 2.4M engagements

"You can get much better results out of ChatGPT by forcing it to go through a step-by-step process. An example: ChatGPT is generally really bad at creating interesting puzzles and scenarios to solve either making things too easy or impossible. But step-by-step approaches work. Prompt: "You a game master. Your job is to come up with interesting challenges for the player to solve. Describe a challenging fantasy scenario and enable me to solve it in an interesting way. You will use the following format to help create a series of responses. Chain of thought: Step 1: Decide on the the scenario"
X Link 2023-04-03T03:54Z 320.8K followers, 2.6M engagements

"It is pretty amazing that a single prompt can have GPT-4 generate ideas select one give the next development steps create a marketing pitch and describe a UX. And one more prompt creates the start of the Python code needed for a rapid prototype. Not perfect but really lowers friction for exploring ideas. First prompt: Please do the following [--] stage process: Stage 1: As a creative idea generator generate [--] diverse ideas for an app for a new type of GPT-4 that increases productivity for writing tasks. make the ideas very creative and aim at narrow specific markets and professions Stage 2:"
X Link 2023-04-08T02:46Z 320.8K followers, 2.1M engagements

"This is quite the paper It gave [--] AI agents motivations & memory and put them in a simulated town. Not only did they engage in complex behavior (including throwing a Valentines Day party) but the actions were rated more human than humans roleplaying. https://arxiv.org/pdf/2304.03442.pdf https://arxiv.org/pdf/2304.03442.pdf"
X Link 2023-04-10T18:51Z 320.8K followers, 5.6M engagements

"Chemists give GPT-4 access to chemical databases & control of off-the-shelf lab robotics to create an "Intelligent Agent system capable of autonomously designing planning & executing complex scientific experiments." They find it exciting. and worrying. https://arxiv.org/ftp/arxiv/papers/2304/2304.05332.pdf https://arxiv.org/ftp/arxiv/papers/2304/2304.05332.pdf"
X Link 2023-04-13T17:27Z 320.8K followers, 2.3M engagements

"The wisdom of the crowd works even if the crowd is totally drunk. This study got undergrads intoxicated & found that while drunk individuals make a lot more errors the consensus of groups of drunk people was as accurate as that of groups of sober people. https://core.ac.uk/download/pdf/42577847.pdf https://core.ac.uk/download/pdf/42577847.pdf"
X Link 2023-04-17T15:06Z 320.8K followers, 1.3M engagements

"The new version of Midjourney that released yesterday shows how far AI has come in making commerical-level images from text alone Here is what you get for "modern outfits inspired by Van Gogh/ Basquiat/ Monet/ Rothko fashion photoshoot" Each one is the first try no revisions"
X Link 2023-05-03T18:21Z 320.8K followers, 2M engagements

"Everyone freaked out for years that violent video games made kids violent but it turns out that they probably dont have any big effects. The consensus of most large studies of violent games (using real game play data not lab tests) shows no big impact on real-life aggression"
X Link 2023-06-22T00:14Z 320.8K followers, 2.6M engagements

"In this study AI was more accurate than two thirds of radiologists yet when radiologists had AI help their diagnoses did not improve. Why Humans ignored the AIs advice when it conflicted with their views. A big barrier to future human-AI collaboration https://blueprintcdn.com/wp-content/uploads/2023/07/Blueprint-Discussion-Paper-2023.10-Agarwal-Moehring-Rajpurkar-Salz_2.pdf https://blueprintcdn.com/wp-content/uploads/2023/07/Blueprint-Discussion-Paper-2023.10-Agarwal-Moehring-Rajpurkar-Salz_2.pdf"
X Link 2023-08-01T00:44Z 320.8K followers, 819.1K engagements

"In a new paper showing that AI comes up with more effective prompts for other AIs than humans do there is this gem that shows how weird AIs are. The single most effective prompt was to start by telling the AI "Take a deep breath and work step-by-step" https://arxiv.org/pdf/2309.03409.pdf https://arxiv.org/pdf/2309.03409.pdf"
X Link 2023-09-08T18:00Z 320.8K followers, 1.2M engagements

"Nice analogy. We are at the KT Boundary for information. Archivists should lock down the pre-2023 information world. What comes after is going to be different. @emollick Like radiocarbon dating pre/post nuclear era. @emollick Like radiocarbon dating pre/post nuclear era"
X Link 2023-11-27T01:10Z 320.8K followers, 303.2K engagements

"This can hack AI now. _ (__/) () / Paper showing that ASCII art can get around AI guardrails. Its the return of 1980s hackers. https://arxiv.org/pdf/2402.11753.pdf https://arxiv.org/pdf/2402.11753.pdf"
X Link 2024-03-01T22:08Z 320.8K followers, 2M engagements

"The speed and extra coding oomph of GPT-4o make it really powerful at analysis compared to GPT-4. Analyze this. Visualize it. Do sophisticated analysis Given a dataset of superheroes and no other context it does really impressive visualization PCA clustering analysis"
X Link 2024-05-13T20:08Z 320.8K followers, 1.7M engagements

"Civilization VII was announced today. The Civ series is sort of like Enders Game but for management rather than murdering aliens. Business school students who were good at Civ V also turn out to be better planners organizers and problem-solvers in this small experiment"
X Link 2024-06-08T01:05Z 320.8K followers, 1.7M engagements

"👀Claude handles an insane request: Remove the squid The document appears to be the full text of the novel "All Quiet on the Western Front" by Erich Maria Remarque. It doesn't contain any mention of squid that I can see. Figure out a way to remove the 🦑"
X Link 2024-07-18T01:50Z 320.8K followers, 1.5M engagements

"Wow diffusion models (used in AI image generation) are also game engines - a type of world simulation. By predicting the next frame of the classic shooter DOOM you get a playable game at [--] fps without any underlying real game engine. This video is from the diffusion model"
X Link 2024-08-28T04:16Z 320.8K followers, 1.8M engagements

"There are now papers showing that the migration of academic communities from Twitter to other social media sites mostly failed. But they never fully returned to Twitter either. I wonder if we will ever see so many different communities interacting in the same online space again"
X Link 2024-09-08T22:10Z 320.8K followers, 1.2M engagements

"Google's NotebookLM is the current best "wow this is amazing & useful" demo of AI Here I gave it the entire text of my book it turned it into a podcast a study guide FAQ timeline & quite accurate chat Listen to the first few minutes of the "podcast." Seriously just listen"
X Link 2024-09-18T19:01Z 320.8K followers, 537.1K engagements

"You really really should not trust audio clips anymore Even a couple months ago it used to take a commercial service to clone a voice. No more. Here is me creating a voice clone of myself using just a [--] second reference clip on my home computer This is all real time no cuts"
X Link 2024-10-14T00:16Z 320.8K followers, 616K engagements

"Half the nitrogen in your body comes from the Haber-Bosch process. This is the most important scientist in human history and its not even close (imo) https://t.co/GqXGi8Zfs4 This is the most important scientist in human history and its not even close (imo) https://t.co/GqXGi8Zfs4"
X Link 2024-10-19T22:03Z 320.8K followers, 346.2K engagements

""Hey Claude with computer use watch this construction site video & write up things you see that dangerous or good create a spreadsheet of critical issues to address" (sped up) How firms use AI as manager coach or panopticon is going to have a big impact on what work becomes"
X Link 2024-11-04T01:58Z 320.8K followers, 841.8K engagements

"It always stuck me as interesting that the very first things people felt comfortable giving over to AI to do for them were the most intimate things: birthday cards wedding speeches childrens stories eulogies (yes really). On the train next to me someone is transcribing text ChatGPT wrote on to a handwritten card https://t.co/euysLJkF2Y On the train next to me someone is transcribing text ChatGPT wrote on to a handwritten card https://t.co/euysLJkF2Y"
X Link 2025-02-08T22:22Z 320.8K followers, 552.7K engagements

"Very cool example of how to deploy generative AI in gaming by Lennart Meincke creating a new texture for a car in GTA IV with a single prompt in GPT-4o (and an example of an existing texture). Going from image creation to in-game use in a couple minutes"
X Link 2025-03-27T02:22Z 320.8K followers, 231.2K engagements

"Huh. Looks like Plato was right. A new paper shows all language models converge on the same "universal geometry" of meaning. Researchers can translate between ANY model's embeddings without seeing the original text. Implications for philosophy and vector databases alike"
X Link 2025-05-23T02:44Z 320.8K followers, 1.7M engagements

"I am starting to think sycophancy is going to be a bigger problem than pure hallucination as LLMs improve. Models that wont tell you directly when you are wrong (and justify your correctness) are ultimately more dangerous to decision-making than models that are sometimes wrong"
X Link 2025-07-13T22:10Z 320.8K followers, 385.7K engagements

"It seems like there is not enough of a policy response to the fact that with 57M miles of data Waymos autonomous vehicles experience 85% less serious injuries & 79% less injuries overall than cars with human drivers. [---] million are injured & 40k killed in US accidents a year"
X Link 2025-08-23T13:41Z 320.8K followers, 1.2M engagements

"This paper shows that you can predict actual purchase intent (90% accuracy) by asking an LLM to impersonate a customer with a demographic profile giving it a product & having it give its impressions which another AI rates. No fine-tuning or training & beats classic ML methods"
X Link 2025-10-10T14:31Z 320.8K followers, 917.6K engagements

"AI resources I work on that might be useful: My NY Times bestseller Co-Intelligence (now in [--] languages): The Generative AI Lab at Wharton (free prompts & research): OneUsefulThing my free newsletter: https://www.oneusefulthing.org/ https://gail.wharton.upenn.edu/research-and-insights/ https://www.penguinrandomhouse.com/books/741805/co-intelligence-by-ethan-mollick/ https://www.oneusefulthing.org/ https://gail.wharton.upenn.edu/research-and-insights/ https://www.penguinrandomhouse.com/books/741805/co-intelligence-by-ethan-mollick/"
X Link 2025-10-20T00:30Z 320.8K followers, 163K engagements

"Surprisingly good for the first try. Nano banana pro: "create a map of the US where every state is made out of its most famous food (the states should actually look like they are made of the food not a picture of the food). Check carefully to make sure each state is right.""
X Link 2025-12-02T05:05Z 320.8K followers, 257.3K engagements

"Had Claude Code build a little plugin that visualizes the work Claude Code is doing as agents working in an office with agents doing work and passing information to each other. New subagents are hired they acquire skills and they turn in completed work. Fun start"
X Link 2026-01-14T04:05Z 320.8K followers, 465.3K engagements

"The Claude Constitution shows where Anthropic thinks this is all going. It is a massive document covering many philosophical issues. I think it is worth serious attention beyond the usual AI-adjacent commentators. Other labs should be similarly explicit. https://www.anthropic.com/constitution https://www.anthropic.com/constitution"
X Link 2026-01-21T18:28Z 320.8K followers, 201K engagements

"Everyone is starting to sound like AI even in spoken language Analysis of [------] transcripts of videos of talks & presentations from academic channels finds they increasingly used words that are favorites of ChatGPT Model collapse except for humans https://arxiv.org/pdf/2409.01754v1 https://arxiv.org/pdf/2409.01754v1"
X Link 2026-01-22T21:07Z 320.8K followers, 390.8K engagements

"Claude in Excel is really good. Its weird that using Microsoft's own Excel agent using Claude [---] often yields weaker answers It seems to be because the Excel agent relies on Excel alone (VLOOKUPs etc) while Claude in Excel does its own analysis and uses Excel for output. https://twitter.com/i/web/status/2014891787051999566 https://twitter.com/i/web/status/2014891787051999566"
X Link 2026-01-24T02:43Z 320.8K followers, 432.7K engagements

"Somewhat surprising that OpenAI has been slower to pivot into non-coding work than Anthropic. Currently there is no OpenAI equivalent to Cowork or Claude for Excel or Claude for Finance despite the fact that the GPT5.2 Thinking/Pro are among the best agentic models right now"
X Link 2026-02-03T01:57Z 320.8K followers, 35.4K engagements

"The Opus [---] system card has some extremely wild stuff that remind you about how weird a technology this is. These paragraphs are really worth reading"
X Link 2026-02-06T00:40Z 320.8K followers, 211.9K engagements

"If you are considering taking a job offer you may want to ask what your token budget will be"
X Link 2026-02-06T03:56Z 320.8K followers, 527K engagements

"AI labs need to realize their routers are very bad at understanding how hard a problems is in advance when they involve any sort of agentic work and especially the knowledge work that they are all focused on helping with. Paid accounts need to let humans pick effort levels"
X Link 2026-02-08T17:35Z 320.8K followers, 21.2K engagements

"If you are in any situation where being right matters you would at this point be making a mistake to not ask a frontier LLM for help. That can mean checking your own work second opinions on other experts or getting help with a complex problem. Have judgement but use them From a recent conversation: I'm already at the point where I trust LLMs broadly over most human experts - doctors accountants lawyers. LLMs might not be 99th percentile but they're reliably 90th and humans just aren't. The crossover was sometime in early [----] I think. From a recent conversation: I'm already at the point where"
X Link 2026-02-10T02:45Z 320.8K followers, 61.9K engagements

"LLMs tripled new book releases since [----]. Average quality fell: most new entries are indeed slop BUT books 100-1000 per category are actually better than before & pre-LLM authors got more productive. And since people only read the good books it is net positive for readers"
X Link 2026-02-10T18:17Z 320.8K followers, 34.8K engagements

"The AI Labs don't yet do a good job explaining how the upgrades to their harnesses change work. For example since Opus [---] Claude Code will spontaneously use subagents to do work in parallel. This is very helpful with a real impact on tasks but was sort of quietly rolled out"
X Link 2026-02-10T21:02Z 320.8K followers, 16.6K engagements

"The new ByteDance SeeDance [---] video model is VERY good. This is the very first output from my very first prompt: "A nature documentary about an otter flying an airplane""
X Link 2026-02-11T02:24Z 320.8K followers, 302.3K engagements

"Seedance: "An influencer in a TikTok video wearing an otter baseball cap showing off the weird swirling vortex they have in their living room. Cheese shoots out of the vortex every few seconds forcing them to move around the room" Again very first attempt"
X Link 2026-02-11T03:01Z 320.8K followers, 21.7K engagements

"Same prompt attempt [--] of [--]. Also excellent different vibes (and a creative transition)"
X Link 2026-02-11T17:13Z 320.8K followers, 16.2K engagements

"I think many knowledge workers who spend an hour with Cowork will get that "Claude Code" moment that has been roiling X for the past few weeks"
X Link 2026-02-11T17:38Z 320.8K followers, 18.7K engagements

"@krishnanrohit This organization could have been a series of markdown files"
X Link 2026-02-11T18:14Z 320.8K followers, 32.8K engagements

"The poetry tastes of GenAI: "I want you to suggest two poems that you think apply very well the current state of GenAI models like you. Dont just pick popular poems and back justify. Think hard about options first." ChatGPT Gemini & Claude all suggest Borges's "The Golem""
X Link 2026-02-11T20:07Z 320.8K followers, 14.5K engagements

"Justification. Second choices: Claude loved Rilke's "Archaic Torso of Apollo". ChatGPT [---] Thinking went with "Naming of Parts" and Gemini went with "The Idea of Order at Key West" by Wallace Stevens"
X Link 2026-02-11T20:24Z 320.8K followers, [----] engagements

"Does Claude Cowork support agents I know it launches its own subagents but can you import agents from Claude Code into Cowork For example I have an Code image creation agent that uses the GPT-imagegen API I want to do image generation in Cowork does anyone know how"
X Link 2026-02-12T03:42Z 320.8K followers, 26.2K engagements

"There are reasons formal education could become more important as AI gets better. Learning through informal apprenticeship (entry level white collar work) is being undermined by AI as is ability to signal early work quality. We are pretty good at both those things in higher ed"
X Link 2026-02-12T15:07Z 320.8K followers, 17.9K engagements

"We dont have any good barriers to bots flooding every online space. They can use browsers and make payments. In fact it is arguably a better deal to pay for a bot to access a social media site since it can post and carry out your agenda when you sleep. I feel like very few people have updated from the experience where Twitter added a literal payment to become a verified user and it made the bot problem at best equally bad but mostly worse https://t.co/YUq3GOGboN I feel like very few people have updated from the experience where Twitter added a literal payment to become a verified user and it"
X Link 2026-02-12T17:04Z 320.8K followers, 13.6K engagements

"Less than a year later Lennart (a well-known GTA modder) rebuilt this with Nano Banana Pro which is good enough that it can generate GTA car textures with no need for human adjustments. He also made a Claude skill that automatically builds deploys & tests them in-game. Very cool example of how to deploy generative AI in gaming by Lennart Meincke creating a new texture for a car in GTA IV with a single prompt in GPT-4o (and an example of an existing texture). Going from image creation to in-game use in a couple minutes. https://t.co/lhe3oNxqXL Very cool example of how to deploy generative AI"
X Link 2026-02-12T20:08Z 320.8K followers, 191.5K engagements

"The Gemini models are really good but the Gemini interface isn't built for agentic work. It doesn't have the same full harness of tools as Claude and ChatGPT and it doesn't do a good job explaining what it did and why and it is unable to output files or other useful artifacts"
X Link 2026-02-12T21:16Z 320.8K followers, 15.4K engagements

"I do not think this is true speaking as someone who spends a lot of time researching AI & work. However I do not think that the opposite is true either AI will have start to have a big impact on white collar work during the next [--] months. But it will not be fully automated CEO of Microsoft AI Mustafa Suleyman joins FT editor Roula Khalaf to explain why most of the tasks accountants lawyers and other professionals currently undertake will be fully automated by AI within the next [--] to [--] months https://t.co/yYKzS7NIOP https://t.co/HvA6Q7KgIc CEO of Microsoft AI Mustafa Suleyman joins FT"
X Link 2026-02-13T00:33Z 320.8K followers, 26.6K engagements

"Rather than converting everything into random benchmark numbers that humans have no comparisons for we should just use D&D stat blocks for AI (CHR INT WIS) and robots (STR DEX CON). You then can do reasonable comparisons like whether any given AI can outsmart a beholder"
X Link 2026-02-13T01:21Z 320.8K followers, 10.2K engagements

"I am surprised that we dont see more governments and non-profits going all-in on transformational AI use cases for good. There are areas like journalism & education where funding ambitious civic-minded & context-sensitive moonshots could make a difference and empower people. I continue to think Claude Code is pretty impressive at producing informative data-based local journalism at low cost with minimal human involvement. A skilled editor and a couple of real reporters working with this tool could do a lot. I continue to think Claude Code is pretty impressive at producing informative"
X Link 2026-02-13T02:05Z 320.8K followers, 27.8K engagements

"One of the most impactful things that OpenAI could do to make more people understand how good AI has become is to TELL THEM when they are given to a weak model. The router is incredibly confusing to almost everyone and people dont get that there is no such thing as ChatGPT-5.2 A big mistake by AI companies: to do this I'd either a) ask decent models to write code or b) use something like [---] Pro which does this by default. But Holly mentions she used "GPT 5.2" - can you fault her for not knowing that refers to [--] models of wildly varying quality A big mistake by AI companies: to do this I'd"
X Link 2026-02-13T03:11Z 320.8K followers, 35.2K engagements

"To give credit to the economists these two papers from early [----] (by @danielrock @robseamans and others who I am not sure are still on X) did a great job forecasting which jobs would turn out to be most exposed to AI using O*NET and they were written during the GPT-3.5 era"
X Link 2026-02-13T04:06Z 320.8K followers, 49.4K engagements

"Very hard to find AI benchmarks that don't look like this (and seems unlikely that a lot of these benchmarks are important enough to actually train on)"
X Link 2026-02-13T19:03Z 320.8K followers, 23.4K engagements

"Experimental evidence for The Whispering Earring (though this could be just seen as the already well-documented phenomenon of automation bias in a new form) https://papers.ssrn.com/sol3/papers.cfmabstract_id=6097646 https://papers.ssrn.com/sol3/papers.cfmabstract_id=6097646"
X Link 2026-02-14T00:39Z 320.8K followers, 27.6K engagements

""Don't read the replies" has taken on an entirely different context as the replies are now from AIs who write meaning-shaped comments that you actually have to spend a split second thinking about. Gets around the defenses of those of us used to filtering bad & stupid comments"
X Link 2026-02-14T00:44Z 320.8K followers, 13.2K engagements

"Impressive benchmarks for the new Chinese LLM. The system card notes some gaps with US closed source models in code generation & wide knowledge so be interested to see it in operation. Not clear it is open weights though If not represents a large shift in the AI market. Seed [---] is finally out 🔥 https://t.co/XXPqBSaE0E Seed [---] is finally out 🔥 https://t.co/XXPqBSaE0E"
X Link 2026-02-14T13:06Z 320.8K followers, 26.4K engagements

"(Large shift because the Chinese AI approach has generally been releasing models for free rather than competing with US closed source models if that isnt happening changes the economic pressures on OpenAI Anthropic etc)"
X Link 2026-02-14T13:09Z 320.8K followers, 10.7K engagements

"I hate these LLM comments (though you are okay lone human). They used to at least understand the post and restate it. Now they get basic facts wrong. Use bigger models and better prompts"
X Link 2026-02-14T13:38Z 320.8K followers, [----] engagements

"The transition from AI cant do novel science to of course AI does novel science will be like every other similar AI transition. First the over-enthusiastic claims then smart people use AI to help them then AI starts to do more of the work then minor discoveries & then"
X Link 2026-02-14T14:17Z 320.8K followers, 39.3K engagements

"One of the things to watch out for as AI advances is the verifiably becomes something only a vanishingly small number of people can do (below is a mathematician on AI proofs) We need to start thinking harder about that problem (multiple AIs working together something else) @ben_golub I think no. Ive pinged some experts about [--] (fwiw Im fairly confident its wrong) but havent heard back yet. AFAIK nothing is really checked; the only other things Im confident about is [--] is essentially right and [--] is very wrong. Solution to [--] seems to be credible but ofc I @ben_golub I think no. Ive pinged some"
X Link 2026-02-14T22:06Z 320.8K followers, 39.8K engagements

"If you know a field well you can probably edit or write your own skill which since it can be run many times by many people can have a large impact. And they are written in plain language that a non-technical expert can compose test & scale. Will be valuable for organizations"
X Link 2026-02-15T19:16Z 320.8K followers, 10.2K engagements

"The crazy part is that the AI Labs have generally been right. Like the stuff they hyped in [----] turned out to be real and working today. That doesn't mean that the stuff they are predicting for [----] will also be real but it is probably worth noting those predictions & watching"
X Link 2026-02-16T04:45Z 320.8K followers, 27.2K engagements

"New randomized controlled trial of students using GPT-4 as a tutor in Nigeria. [--] weeks of after-school AI tutoring = [--] years of typical learning gains outperforming 80% of other educational interventions. And it helped all students especially girls who were initially behind"
X Link 2025-01-15T20:55Z 320.8K followers, 4.3M engagements

"The thing about Moltbook (the social media site for AI agents) is that it is creating a shared fictional context for a bunch of AIs. Coordinated storylines are going to result in some very weird outcomes and it will be hard to separate "real" stuff from AI roleplaying personas"
X Link 2026-01-30T16:57Z 320.8K followers, 126.6K engagements

"There is no sign that Democrats or Republicans have different propensities to use AI: "the politics of AI is not primarily driven by ideological resistance or enthusiasm for the technology but rather by structural differences in where people work and what skills they possess." What do you know Political affiliation has no impact on whether you use AI. Why would we think it would https://t.co/EC5ZFLd3Hz What do you know Political affiliation has no impact on whether you use AI. Why would we think it would https://t.co/EC5ZFLd3Hz"
X Link 2026-02-09T15:33Z 320.8K followers, 16.9K engagements

"So far telling a satisfying and well-written medium-length story has proved far harder for LLMs than mathematical proofs music generation research reports code and many other forms of work. The technical reasons are pretty clear but they are supposed to be language models"
X Link 2026-02-09T22:49Z 320.8K followers, 123.9K engagements

"This might be the first hot take on how technology tells us how to live our lives destroying our ability to make human decisions. The technology in question is the sundial. From a 3rd century BCE Roman adaptation of a Greek play as discussed in Kerrs The Ordered Day"
X Link 2026-02-10T05:51Z 320.8K followers, 95.9K engagements

"ByteDance's Seedance 2.0: "Monica's apartment from the show Friends except all of the friends are otters wearing wigs. The otter with a Rachel wig says "Is anything weird" and the one with a Joey wig says "Nope all is normal"" Huh"
X Link 2026-02-11T02:28Z 320.8K followers, 25.2K engagements

"SeeDance 2.0: "An anime where an otter goes into a large mech with lots of quick shots of mechanical parts and gears turning. The otter gives a grim thumbs up and then pilots the mech flying into battle against an octopus made of marble." Again this was the very first try"
X Link 2026-02-11T02:33Z 320.8K followers, 552.6K engagements

"Seedance: "A documentary about how otters view Ethan Mollick's "Otter Test" which judges AIs by their ability to create images of otters sitting in planes" Again first result"
X Link 2026-02-11T03:26Z 320.8K followers, 63.7K engagements

"Example of Seedance with some consistency issues but still: "Action sequence shot for a big budget action movie where two elegantly dressed woman on giant snails race slowly around a track as gunners on the snails fire at each other. Lots of quick cuts and action movie cliches""
X Link 2026-02-11T03:54Z 320.8K followers, 52.5K engagements

"I have noticed that compared to other models there is not a lot of diversity in main characters somewhat surprising that a model from China hasn't produced any Asian actors in any of these scenes. (though I haven't made that many videos so my view is still limited)"
X Link 2026-02-11T03:57Z 320.8K followers, [----] engagements

"Second attempt is also great. I am especially impressed that they keep getting otter "emotions" right (or at least how humans would interpret them) and they have a satisfying storyline for a [--] second clip rather than just cutting off"
X Link 2026-02-11T16:53Z 320.8K followers, [----] engagements

"I pointed Claude Cowork at a set of [---] documents (PPTs Word docs Excel) that were initially hand-created for my class at Wharton & expanded on by AI. They make up a very complex business case with lots of issues & opportunities AI was able to one-shot the case from documents"
X Link 2026-02-11T17:34Z 320.8K followers, 131.6K engagements

"Its a weird time to post about AI because a lot of people are vastly underestimating what AI can do & how many large-scale impacts on work are inevitable with todays models while a lot of other people underestimate the real world problems involved in getting value from AI"
X Link 2026-02-11T18:09Z 320.8K followers, 38K engagements

"So maybe the argument that the world is compute constrained and there arent anywhere enough cheap tokens to go around wasnt so fanciful after all"
X Link 2026-02-11T18:40Z 320.8K followers, 24.4K engagements

"Rilke was a good choice"
X Link 2026-02-11T20:51Z 320.8K followers, [----] engagements

"Less than a year from announcement to near saturation. (On to ARC-AGI-3) Unlike ARC-AGI-1 this new version is not easily brute-forced. Current top AI approaches score 0-4%. All base LLMs (GPT-4.5 Claude [---] Sonnet Gemini [--] etc.) score 0%. Single-CoT reasoning models (Claude Thinking R1 o3-mini) score 0-1%. So you can't solve these tasks via https://t.co/wjN6t0XnLi Unlike ARC-AGI-1 this new version is not easily brute-forced. Current top AI approaches score 0-4%. All base LLMs (GPT-4.5 Claude [---] Sonnet Gemini [--] etc.) score 0%. Single-CoT reasoning models (Claude Thinking R1 o3-mini) score"
X Link 2026-02-12T17:11Z 320.8K followers, 107.4K engagements

"The most interesting thing about Seedance [---] is that clips can be just long enough (15 seconds) to have something interesting happen and the LLM behind it is good enough to actually make a little narrative arc rather than cut off the way Veo and Sora do. Changes the impact. Seedance: "A documentary about how otters view Ethan Mollick's "Otter Test" which judges AIs by their ability to create images of otters sitting in planes" Again first result. https://t.co/jbKQ765ZHs Seedance: "A documentary about how otters view Ethan Mollick's "Otter Test" which judges AIs by their ability to create"
X Link 2026-02-12T20:51Z 320.8K followers, 23.2K engagements

"Waiting for a Gemini Deep Think result a ChatGPT [---] Pro result and a Claude Code task at the same time. I think I unlocked some sort of AI Steam achievement"
X Link 2026-02-12T20:53Z 320.8K followers, 22.2K engagements

"The new Deep Think is very very smart but is constrained by the Gemini interface compared to GPT-5.2 Pro. If I can't see actual work in the thinking trace or get downloadable files as proof of work or see evidence of the code & statistics it applied checking results is hard"
X Link 2026-02-12T21:13Z 320.8K followers, 53.8K engagements

Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing