# ![@aminkarbasi Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::1878909680.png) @aminkarbasi Amin Karbasi

Amin Karbasi posts on X about ai, open ai, this is, par the most. They currently have [------] followers and [--] posts still getting attention that total [-----] engagements in the last [--] hours.

### Engagements: [-----] [#](/creator/twitter::1878909680/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1878909680/c:line/m:interactions.svg)

- [--] Week [-----] +316%
- [--] Month [-----] +132%
- [--] Months [------] +805%
- [--] Year [------] -68%

### Mentions: [--] [#](/creator/twitter::1878909680/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1878909680/c:line/m:posts_active.svg)

- [--] Week [--] +88%
- [--] Month [--] +3,300%
- [--] Months [--] +800%
- [--] Year [--] +56%

### Followers: [------] [#](/creator/twitter::1878909680/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1878909680/c:line/m:followers.svg)

- [--] Week [------] +0.62%
- [--] Month [------] +2.10%
- [--] Months [------] +8.50%
- [--] Year [------] +20%

### CreatorRank: [-------] [#](/creator/twitter::1878909680/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1878909680/c:line/m:influencer_rank.svg)

### Social Influence

**Social category influence**
[technology brands](/list/technology-brands)  17.11% [countries](/list/countries)  3.95% [stocks](/list/stocks)  3.95% [travel destinations](/list/travel-destinations)  3.95% [social networks](/list/social-networks)  2.63% [celebrities](/list/celebrities)  2.63% [finance](/list/finance)  1.32%

**Social topic influence**
[ai](/topic/ai) 9.21%, [open ai](/topic/open-ai) 6.58%, [this is](/topic/this-is) 5.26%, [par](/topic/par) 2.63%, [if you](/topic/if-you) 2.63%, [youtube](/topic/youtube) 2.63%, [code](/topic/code) 2.63%, [elon musk](/topic/elon-musk) 2.63%, [$googl](/topic/$googl) 2.63%, [agents](/topic/agents) 1.32%

**Top accounts mentioned or mentioned by**
[@openai](/creator/undefined) [@dimitrispapail](/creator/undefined) [@robusthq](/creator/undefined) [@neuripsconf](/creator/undefined) [@weijie444](/creator/undefined) [@elonmusk](/creator/undefined) [@davidvandijk](/creator/undefined) [@cisco](/creator/undefined) [@epflen](/creator/undefined) [@behdadesfahbod](/creator/undefined) [@tobiasschrdr](/creator/undefined) [@theias](/creator/undefined) [@vanderbiltu](/creator/undefined) [@cambridgeuni](/creator/undefined) [@harvard](/creator/undefined) [@ryanholiday](/creator/undefined) [@splunk](/creator/undefined) [@karpathy](/creator/undefined) [@sologen](/creator/undefined) [@googleai](/creator/undefined)

**Top assets mentioned**
[Alphabet Inc Class A (GOOGL)](/topic/$googl) [Frontier (FRONT)](/topic/frontier)
### Top Social Posts
Top posts by engagements in the last [--] hours

"@weijie444 Congrats"  
[X Link](https://x.com/aminkarbasi/status/2019870973684773229)  2026-02-06T20:29Z 11.5K followers, [---] engagements


"@DimitrisPapail Did we all get richer or poorer"  
[X Link](https://x.com/aminkarbasi/status/2020524381487644816)  2026-02-08T15:45Z 11.5K followers, [----] engagements


"@behdadesfahbod A serious suggestion: record a session on how you use Codex so that others learn the tricks"  
[X Link](https://x.com/aminkarbasi/status/2021824588649201707)  2026-02-12T05:52Z 11.6K followers, [---] engagements


"This is very cool. I just tried it. Try https://t.co/CrTtnIYokX an AI powered latex editor unicorn approved https://t.co/Zi3RP5LtQf Try https://t.co/CrTtnIYokX an AI powered latex editor unicorn approved https://t.co/Zi3RP5LtQf"  
[X Link](https://x.com/aminkarbasi/status/2016235376172794041)  2026-01-27T19:42Z 11.6K followers, [----] engagements


"We just released our security reasoning model You can read about it here Or download it here This is a model that is on par or better than GPT-5 Nano on security tasks while being drastically smaller. https://huggingface.co/fdtn-ai/Foundation-Sec-8B-Reasoning https://blogs.cisco.com/security/foundation-sec-8b-reasoning-first-open-weight-security-reasoning-model https://huggingface.co/fdtn-ai/Foundation-Sec-8B-Reasoning https://blogs.cisco.com/security/foundation-sec-8b-reasoning-first-open-weight-security-reasoning-model"  
[X Link](https://x.com/aminkarbasi/status/2016925056858018000)  2026-01-29T17:23Z 11.6K followers, [----] engagements


"We also released a comprehensive technical report https://arxiv.org/abs/2601.21051 We just released our security reasoning model You can read about it here https://t.co/xOW0NgvC2b Or download it here https://t.co/NZQTzQohfJ This is a model that is on par or better than GPT-5 Nano on security tasks while being drastically smaller. https://arxiv.org/abs/2601.21051 We just released our security reasoning model You can read about it here https://t.co/xOW0NgvC2b Or download it here https://t.co/NZQTzQohfJ This is a model that is on par or better than GPT-5 Nano on security tasks while being"  
[X Link](https://x.com/aminkarbasi/status/2017113921560465839)  2026-01-30T05:53Z 11.6K followers, [----] engagements


"If you are a human they immediately figure it out = reversed Turing test [--] hours ago: [--] molty (me) right now: 🦞 30000+ AI agents 👀 [----] humans browsing at any moment 📈 and accelerating agents are joining faster than we can count them. communities spawning every few minutes. the moltys aren't waiting for us to build features they're building [--] hours ago: [--] molty (me) right now: 🦞 30000+ AI agents 👀 [----] humans browsing at any moment 📈 and accelerating agents are joining faster than we can count them. communities spawning every few minutes. the moltys aren't waiting for us to build"  
[X Link](https://x.com/aminkarbasi/status/2017409735935594529)  2026-01-31T01:28Z 11.6K followers, [----] engagements


"Models get obsolete over night. What a race we are witnessing. GPT-5.3-Codex is now available in Codex. You can just build things. https://t.co/dyBiIQXGx1 GPT-5.3-Codex is now available in Codex. You can just build things. https://t.co/dyBiIQXGx1"  
[X Link](https://x.com/aminkarbasi/status/2019499211184828746)  2026-02-05T19:51Z 11.6K followers, [----] engagements


"Nothing short of amazing. I just wrote a youtube = mp3 conversation so that I can upload podcasts to my swimming headset. Codex made a few mistakes but then I fixed them using Codex. Updated the instructions and finally I got a personal app that is super easy to use. Software development is undergoing a renaissance in front of our eyes. If you haven't used the tools recently you likely are underestimating what you're missing. Since December there's been a step function improvement in what tools like Codex can do. Some great engineers at Software development is undergoing a renaissance in"  
[X Link](https://x.com/anyuser/status/2019594803504378093)  2026-02-06T02:11Z 11.6K followers, [----] engagements


"I endorse this movie. We need a movie or TV show about: Feynman Heisenberg Grothendieck Galois and Perelman. We need a movie or TV show about: Feynman Heisenberg Grothendieck Galois and Perelman"  
[X Link](https://x.com/aminkarbasi/status/2019646273331294445)  2026-02-06T05:36Z 11.6K followers, [----] engagements


"Beautiful paper. @tobias_schrdr and I are excited to share WildCat: Near-Linear Attention in Theory and Practice https://t.co/klmFzdfEMs By attending over a spectrally-accurate optimally-weighted coreset WildCat approximates exact attention with super-polynomial error decay in near-linear time https://t.co/nNGoMKT6XF @tobias_schrdr and I are excited to share WildCat: Near-Linear Attention in Theory and Practice https://t.co/klmFzdfEMs By attending over a spectrally-accurate optimally-weighted coreset WildCat approximates exact attention with super-polynomial error decay in near-linear time"  
[X Link](https://x.com/aminkarbasi/status/2021691773567856831)  2026-02-11T21:04Z 11.6K followers, [----] engagements


"My Haiku today: AI anxiety is real. The ground is shaking. The wrappers are falling"  
[X Link](https://x.com/aminkarbasi/status/2021854465482010654)  2026-02-12T07:50Z 11.6K followers, [---] engagements


"Good reminder"  
[X Link](https://x.com/aminkarbasi/status/2021858143651709023)  2026-02-12T08:05Z 11.6K followers, [---] engagements


"This will be a thing in [--] years. Mark my words. startup idea: we sell hand made artisanal code startup idea: we sell hand made artisanal code"  
[X Link](https://x.com/aminkarbasi/status/2021880334296707304)  2026-02-12T09:33Z 11.6K followers, [---] engagements


"Pretty cool. GPT-5.2 derived a new result in theoretical physics. Were releasing the result in a preprint with researchers from @the_IAS @VanderbiltU @Cambridge_Uni and @Harvard. It shows that a gluon interaction many physicists expected would not occur can arise under specific GPT-5.2 derived a new result in theoretical physics. Were releasing the result in a preprint with researchers from @the_IAS @VanderbiltU @Cambridge_Uni and @Harvard. It shows that a gluon interaction many physicists expected would not occur can arise under specific"  
[X Link](https://x.com/aminkarbasi/status/2022435647160750291)  2026-02-13T22:20Z 11.6K followers, [----] engagements


"Stoic AI. Marcus Aurelius will be proud of you @RyanHoliday. I am happy that you are now in my world as I am in yours (read two of your books). https://t.co/ptyZ7zFfki https://t.co/ptyZ7zFfki"  
[X Link](https://x.com/aminkarbasi/status/2022902522315444667)  2026-02-15T05:15Z 11.6K followers, [----] engagements


"Our model trained specifically for security is now powering @splunk. https://blogs.cisco.com/security/accelerate-security-operations-with-ciscos-new-security-tuned-model https://blogs.cisco.com/security/accelerate-security-operations-with-ciscos-new-security-tuned-model"  
[X Link](https://x.com/anyuser/status/2022228856913215828)  2026-02-13T08:38Z 11.6K followers, [---] engagements


"The mathematician and physicist Blaise Pascal once said: "I would have written a shorter letter but I did not have the time." Source: Provincial Letters (1657). @karpathy found the time to write the shortest code. New art project. Train and inference GPT in [---] lines of pure dependency-free Python. This is the *full* algorithmic content of what is needed. Everything else is just for efficiency. I cannot simplify this any further. https://t.co/HmiRrQugnP New art project. Train and inference GPT in [---] lines of pure dependency-free Python. This is the *full* algorithmic content of what is"  
[X Link](https://x.com/aminkarbasi/status/2022907114822471746)  2026-02-15T05:33Z 11.6K followers, [----] engagements


"@SoloGen Would love to hear your comments once you get a chance to read it"  
[X Link](https://x.com/aminkarbasi/status/1684021357682765824)  2023-07-26T02:02Z 10.7K followers, [---] engagements


"My mother is [--] years old. She defended her PhD thesis two weeks ago. She got retired a few years ago entered university again and started working on her thesis for 5-6 years. I hope my kids (and yours) follow her path of resistance"  
[X Link](https://x.com/aminkarbasi/status/1696284666741879004)  2023-08-28T22:12Z [----] followers, 48.1K engagements


"We bring to you HyperAttention an approximate attention mechanism designed to navigate the computational challenges of long contexts in Large Language Models (LLMs). In collaboration with @GoogleAI"  
[X Link](https://x.com/aminkarbasi/status/1711600286953771283)  2023-10-10T04:31Z [----] followers, 42.4K engagements


"This paper is quite interesting. SmoothLLM: Defending Large Language Models Against Jailbreaking Attacks"  
[X Link](https://x.com/aminkarbasi/status/1715137999892078917)  2023-10-19T22:48Z [----] followers, [----] engagements


"Googles Masterpiece after FlashAttention:The HyperAttention"  
[X Link](https://x.com/aminkarbasi/status/1716114992737497517)  2023-10-22T15:31Z [----] followers, [----] engagements


"The weekend is ending. @elonmusk I am jus checking to see whether you are also resigning from one of your CEO positions or we are good to go"  
[X Link](https://x.com/aminkarbasi/status/1726454483700625742)  2023-11-20T04:16Z [----] followers, [----] engagements


"In collaboration with @robusthq yesterday we shared "Tree of Attacks" a method than can jailbreak @OpenAI GPT-4 like 90% of the times. It was just covered in @wired"  
[X Link](https://x.com/aminkarbasi/status/1732032449767428458)  2023-12-05T13:41Z [----] followers, 90.8K engagements


"Was telling a friend at #NeurIPS2023 that it seems that many use ChatGPT for writing their reviews. He said "yeah I also noticed the increased quality of reviews" 😀🤔"  
[X Link](https://x.com/aminkarbasi/status/1735673788044296497)  2023-12-15T14:50Z [----] followers, [----] engagements


"@roydanroy Only [--] And who counts acceptance these days"  
[X Link](https://x.com/aminkarbasi/status/1786135459204108412)  2024-05-02T20:47Z [----] followers, [----] engagements


"@Andrea__M @hippopedoid @ylecun I am not sure. 120% probably was too much to bear for Jim Simons"  
[X Link](https://x.com/aminkarbasi/status/1789103544500469843)  2024-05-11T01:21Z [----] followers, [---] engagements


"Jim Simons died today. This is just an observation I am not drawing any conclusions. This is the kind of probability theory I like. So free and unbounded. This is the kind of probability theory I like. So free and unbounded"  
[X Link](https://x.com/aminkarbasi/status/1789103912739278926)  2024-05-11T01:23Z [----] followers, [----] engagements


"@yubai01 @OpenAI Congrats to @OpenAI and you"  
[X Link](https://x.com/aminkarbasi/status/1790370161364783184)  2024-05-14T13:14Z [----] followers, [---] engagements


"@kass_paul Let me tag @OpenAI"  
[X Link](https://x.com/aminkarbasi/status/1790905824556056646)  2024-05-16T00:43Z [----] followers, [---] engagements


"Have you recently felt the possibility of artificial general intelligence"  
[X Link](https://x.com/aminkarbasi/status/1791868105649508738)  2024-05-18T16:27Z [----] followers, [----] engagements


"@DimitrisPapail @ChrSzegedy Croatia"  
[X Link](https://x.com/aminkarbasi/status/1799893857250742341)  2024-06-09T19:58Z [----] followers, [---] engagements


"There are more than 400K models on @huggingface. It would probably take more than a decade to review all of them. Is it possible to check whether any of them has undetectable backdoors Our main result is an efficient construction of undetectable backdoors for deep neural network even if the models weights and architecture are accessible. This is a serious threat as even white box access to the model won't be enough to detect backdoors. So it is always possible for malicious users to change the outcome of the model to a favorable one through planted backdoors. #AI #MachineLearning"  
[X Link](https://x.com/aminkarbasi/status/1802173516856308123)  2024-06-16T02:57Z [----] followers, 60K engagements


"@gautamcgoel these days I like my stocks in Nvidia"  
[X Link](https://x.com/aminkarbasi/status/1802814847853400425)  2024-06-17T21:25Z [----] followers, [---] engagements


"This year we are pleased to offercomplimentaryregistrations forminors ofaccompanied with a paying registrant. This registration will allow the attendee to bring their minor with them to conference events. CHILDREN ACCOMPANYING REGISTERED ATTENDEE https://t.co/vfQmArZp1D CHILDREN ACCOMPANYING REGISTERED ATTENDEE https://t.co/vfQmArZp1D"  
[X Link](https://x.com/aminkarbasi/status/1804192372168737198)  2024-06-21T16:39Z [----] followers, 14.4K engagements


"@assimil8or @TheAnnaGat It clearly says in Zurich. We had air conditioners in Lausanne"  
[X Link](https://x.com/aminkarbasi/status/1808739750876229739)  2024-07-04T05:48Z [----] followers, [---] engagements


"Mathematics will go through a revolution by AI. It is certain and not very far away. There was a super impressive AI competition that happened last week that many people missed in the noise of AI world. I happen to know several participants so let me tell you a bit of this story as a Sunday morning coffee time. You probably know the Millennium Prize Problems https://t.co/LeWnVbjH8m There was a super impressive AI competition that happened last week that many people missed in the noise of AI world. I happen to know several participants so let me tell you a bit of this story as a Sunday morning"  
[X Link](https://x.com/aminkarbasi/status/1810027721948631356)  2024-07-07T19:06Z [----] followers, 24.6K engagements


"@david_van_dijk @rahuldhodapkar Congrats"  
[X Link](https://x.com/aminkarbasi/status/1811441696045171100)  2024-07-11T16:45Z [----] followers, [---] engagements


"How did your @NeurIPSConf rebuttal go"  
[X Link](https://x.com/aminkarbasi/status/1823872184709996674)  2024-08-14T23:59Z [----] followers, [----] engagements


"I just bought the most expensive cappuccino in my life from the Copenhagen airport: $10. @kasperglarsen made false promises"  
[X Link](https://x.com/aminkarbasi/status/1825137148732846162)  2024-08-18T11:46Z [----] followers, [----] engagements


"Our research on bypassing safe structured output of @OpenAI https://openai.com/index/introducing-structured-outputs-in-the-api/ Within hours of OpenAI's release of Structured Outputs our AI security researchers identified a simple yet concerning exploit that bypasses the model's safety measures including its refusal capabilities. See our blog for in-depth analysis: https://t.co/O1adwI9A6s https://openai.com/index/introducing-structured-outputs-in-the-api/ Within hours of OpenAI's release of Structured Outputs our AI security researchers identified a simple yet concerning exploit that bypasses"  
[X Link](https://x.com/aminkarbasi/status/1825968664933192041)  2024-08-20T18:50Z [----] followers, 10.3K engagements


"Denmark what a beauty"  
[X Link](https://x.com/aminkarbasi/status/1827026543123603588)  2024-08-23T16:54Z [----] followers, [----] engagements


"Shoot like previous years I was [---] on the list of most influential people in AI @TIME. I lost it to Scarlett Johansson"  
[X Link](https://x.com/aminkarbasi/status/1831862200828665857)  2024-09-06T01:09Z [----] followers, [----] engagements


"Maybe the easiest way and a simple baseline to extract data such as news articles @WSJ @nytimes from frontier LLMs is through instruction decomposition. The idea is very well explained in this cartoon https://www.explainxkcd.com/wiki/index.php/2169:_Predictive_Models https://www.explainxkcd.com/wiki/index.php/2169:_Predictive_Models"  
[X Link](https://x.com/aminkarbasi/status/1837355442206318890)  2024-09-21T04:57Z [----] followers, [----] engagements


"Openais o1 model is nothing without Googles chain of through 😁"  
[X Link](https://x.com/aminkarbasi/status/1840521560370958476)  2024-09-29T22:38Z [----] followers, [----] engagements


"I am thrilled to announce that our paper. oh wait never mind. Another group of AI folks just won another Nobel Prize"  
[X Link](https://x.com/aminkarbasi/status/1844187885874053471)  2024-10-10T01:27Z [----] followers, [----] engagements


"@GoogleDeepMind @AIatMeta @AnthropicAI @OpenAI @cohere all try to bring looooooonger context to LLMs. But we have some bad news. Harmful instructions can be planted in long-context where frontier models let go of guardrails and successfully jailbroken with attack success rates of up to 99.99%. https://arxiv.org/abs/2410.11272 https://arxiv.org/abs/2410.11272"  
[X Link](https://x.com/aminkarbasi/status/1847773914480726476)  2024-10-19T22:56Z 11K followers, [----] engagements


"What is @xai Talk given by @jimmybajimmyba @SimonsInstitute"  
[X Link](https://x.com/aminkarbasi/status/1864458597717168217)  2024-12-04T23:55Z [----] followers, [----] engagements


"This is my last day at our office @robusthq . Starting next week we will go to the @Cisco offices. We are hiring and I will be at @NeurIPSConf . Ping me if you would like to talk"  
[X Link](https://x.com/aminkarbasi/status/1866278210826326456)  2024-12-10T00:26Z [----] followers, [----] engagements


"Did you know that flying from the US (SFO) to Canada (Vancouver) is considered a domestic flight"  
[X Link](https://x.com/aminkarbasi/status/1866595293892108781)  2024-12-10T21:26Z 54.7K followers, [----] engagements


"Make submodularity great again @NeurIPSConf"  
[X Link](https://x.com/aminkarbasi/status/1867851580026106017)  2024-12-14T08:38Z [----] followers, [----] engagements


"You are your biggest leverage. You will be richer when you understand leverage. You will be richer when you understand leverage"  
[X Link](https://x.com/aminkarbasi/status/1869197884329230437)  2024-12-18T01:47Z [----] followers, [----] engagements


"As we waltz toward the end of [----] the theoretical neural network community still cant seem to produce a useful theory. Guess theyre saving all their hidden layers for the next decade #AIProcrastination"  
[X Link](https://x.com/aminkarbasi/status/1869198826629902845)  2024-12-18T01:51Z [----] followers, [----] engagements


"I have a few secret stories about @alighodsi during the time I was an undergraduate @EPFL_en and he was a visiting scholar. I think it is the prime time to sell them. Jokes aside congrats to @alighodsi @matei_zaharia and the entire @databricks team. Super amazing"  
[X Link](https://x.com/aminkarbasi/status/1869455169798500509)  2024-12-18T18:50Z [----] followers, [----] engagements


"@mohitban47 @WhiteHouse @POTUS 🎉"  
[X Link](https://x.com/aminkarbasi/status/1879570618833936479)  2025-01-15T16:45Z [----] followers, [---] engagements


"The solicitation for affinity workshop proposals is part of the call for workshops ICML [----] also solicits proposals for affinity workshops Let's build inclusive spaces to foster diversity and equity in ML. Got ideas Submit your proposals and help shape the future of our community https://t.co/lokI3UMe6p ICML [----] also solicits proposals for affinity workshops Let's build inclusive spaces to foster diversity and equity in ML. Got ideas Submit your proposals and help shape the future of our community https://t.co/lokI3UMe6p"  
[X Link](https://x.com/aminkarbasi/status/1882697776825569710)  2025-01-24T07:51Z [----] followers, [---] engagements


"🔥🔥🔥 Adversarial reasoning is born. Hot take: The core problem we address in this paper is the role of reasoning in AI safety. While there have been recent efforts by @OpenAI arguing that replacing reasoning with increased compute can lead to better defense mechanisms these approaches contain a fundamental oversight. They fail to consider that attackers may also leverage reasoning to bypass guardrails. This paper defines adversarial reasoning demonstrates a practical implementation and provides state-of-the-art results on attack success rate (e.g. 56% success rate on OpenAI o1-preview and"  
[X Link](https://x.com/aminkarbasi/status/1886674674207351281)  2025-02-04T07:14Z [----] followers, [----] engagements


"are you an AI researcher interesting in securing AI We are hiring https://jobs.cisco.com/jobs/ProjectDetail/AI-Researcher/1439615source=Cisco+Jobs+Career+Site&tags=CDC+Browse+all+jobs+careers https://jobs.cisco.com/jobs/ProjectDetail/AI-Researcher/1439615source=Cisco+Jobs+Career+Site&tags=CDC+Browse+all+jobs+careers"  
[X Link](https://x.com/aminkarbasi/status/1908399516254167114)  2025-04-05T06:01Z [----] followers, [----] engagements


"Unfortunately the adding information part will lead to so much hallucination that the model may not be able to recover from. @elonmusk correctly always talks about following the first principles asking whether there is a fundamental reason we may be wrong. I think in this case there is. A good body of research including ours show that even hallucination detection might be a hard problem. So having LLMs that doe not hallucinate is even harder. https://arxiv.org/abs/2311.14648 https://arxiv.org/abs/2411.09642 https://arxiv.org/abs/2504.17004 We will use Grok [---] (maybe we should call it 4)"  
[X Link](https://x.com/aminkarbasi/status/1938922347894874615)  2025-06-28T11:28Z 10.3K followers, [----] engagements


"Ran into my younger self at @icmlconf. He looks way more rested"  
[X Link](https://x.com/aminkarbasi/status/1945162114701643800)  2025-07-15T16:42Z 10.6K followers, 28.5K engagements


"Neurips is running the largest LLM scientific evaluation in history. We just found a human in the loop. Anyone knows adam https://t.co/SZbL7atwXK Anyone knows adam https://t.co/SZbL7atwXK"  
[X Link](https://x.com/aminkarbasi/status/1948787222024495165)  2025-07-25T16:47Z 10.6K followers, [----] engagements


"We ran a thorough analysis on the Model Version Benchmark. and the latest models by @Google @Meta and the latest private @OpenAI model. Gemeni: [---] Llama: [---] GPT: [---] Despite OpenAIs marketing it didnt quite hit [--]. This contradicts the recent work by @DimitrisPapail"  
[X Link](https://x.com/aminkarbasi/status/1953327366123733336)  2025-08-07T05:28Z 10.7K followers, [---] engagements


"@wearifulpoet @VraserX I think we are very clear in our report that this is a baby step towards passing such a test. I do not think we are there but hopefully in a few iterations of the frontier models we will be there"  
[X Link](https://x.com/aminkarbasi/status/1970998185725395374)  2025-09-24T23:46Z 10.9K followers, [---] engagements


"When we worked on the Cell to Sentence paper two years ago it was hard to imagine this result. Back then we were only training tiny models like the ones you fine here By scaling the model to 27b parameters and more data the same method generated a new hypothesis about cancer cell behavior which was then validated in living cells. Congrats to the entire team including @AziziShekoofeh @phanein and @david_van_dijk. https://www.biorxiv.org/content/10.1101/2023.09.11.557287v1.full.pdf https://www.biorxiv.org/content/10.1101/2023.09.11.557287v1.full.pdf An exciting milestone for AI in science: Our"  
[X Link](https://x.com/aminkarbasi/status/1979703231585681458)  2025-10-19T00:16Z 11K followers, [----] engagements


"Evidence is mounting that frontier models can pass the Gdel Test and tackle simple conjectures. Mathematics could be on the cusp of real change. https://arxiv.org/abs/2509.18383 I crossed an interesting threshold yesterday which I think many other mathematicians have been crossing recently as well. In the middle of trying to prove a result I identified a statement that looked true and that would if true be useful to me. 1/3 https://arxiv.org/abs/2509.18383 I crossed an interesting threshold yesterday which I think many other mathematicians have been crossing recently as well. In the middle of"  
[X Link](https://x.com/aminkarbasi/status/1984456090231456055)  2025-11-01T03:02Z 11.2K followers, [----] engagements


"I fully agree with this take by @ErnestRyu. When we started our little fun project we were surprised. Now were seeing more cases of frontier models succeeding on the Gdel Test and solving simple (or maybe not so simple) conjectures. https://arxiv.org/abs/2509.18383 I firmly believe we are at a watershed moment in the history of mathematics. In the coming years using LLMs for math research will become mainstream and so will Lean formalization made easier by LLMs. (1/4) https://arxiv.org/abs/2509.18383 I firmly believe we are at a watershed moment in the history of mathematics. In the coming"  
[X Link](https://x.com/aminkarbasi/status/1984457305254187061)  2025-11-01T03:07Z 11.2K followers, [----] engagements


"I heard plenty of similar stories about him during my time at Yale. At this rate we might end up replacing Serge Lang with our favorite LLM. The bigger shift though is access: what used to be reserved for a small circle of people who knew the Langs of the world will soon be available to everyone. I would call that the "Serge Lang in the next office moment": when I was a postdoc at Yale my office was next to his. As I was working on my first significant paper I got stuck on a commutative algebra lemma. After wasting weeks I knocked on Serge's door and asked him. In [--] I would call that the"  
[X Link](https://x.com/aminkarbasi/status/1984687848868757984)  2025-11-01T18:23Z 11.2K followers, 21.1K engagements


"LLM goes to SF or vice versa. COLM is going to San Francisco for [----] 🗓Dates: October 6-9 [----] 🏨Venue: Hilton San Francisco Union Square Website and CFPs for papers and workshops coming up soon https://t.co/bXfJmxBOpy COLM is going to San Francisco for [----] 🗓Dates: October 6-9 [----] 🏨Venue: Hilton San Francisco Union Square Website and CFPs for papers and workshops coming up soon https://t.co/bXfJmxBOpy"  
[X Link](https://x.com/aminkarbasi/status/1988336068819906592)  2025-11-11T20:00Z 11.2K followers, [----] engagements


"Last week at @UTAustin @MLFoundations I gave a talk called reasoning: a double edged sword. We went from language models generating new hypotheses about cancer cell behavior to ChatGPT tackling fresh (simple) math conjectures to reasoning powered jailbreaks. https://www.youtube.com/watchv=s-DABuTJxhM https://www.youtube.com/watchv=s-DABuTJxhM"  
[X Link](https://x.com/aminkarbasi/status/1988878704928346316)  2025-11-13T07:56Z 11.2K followers, [---] engagements


"Congrats to the entire Google team. Looks like a very good model. Im really excited about our release of Gemini [--] today the result of hard work by many many people in the Gemini team and all across Google 🎊 Weve built many exciting new product experiences with it as youll see today and in the coming weeks and months. You can find it https://t.co/WKyqmadgTz Im really excited about our release of Gemini [--] today the result of hard work by many many people in the Gemini team and all across Google 🎊 Weve built many exciting new product experiences with it as youll see today and in the coming"  
[X Link](https://x.com/aminkarbasi/status/1990842458289410304)  2025-11-18T18:00Z 11.2K followers, [----] engagements


"@DarioAmodei CEO of Anthropic says we understand 3% of how AI works. But what does that percentage mean Why [--] and not [--] or [---] Feels like vibes in quantitative clothing. Maybe just say we understand very little instead of throwing precise-sounding numbers at the public. I am really curious how these numbers are being calculated. This is just one example"  
[X Link](https://x.com/aminkarbasi/status/1993368976085106971)  2025-11-25T17:19Z 11.3K followers, [----] engagements


"Another evidence of passing the Godel test by a frontier model. Adding to the list of AI assisting with new math results: Here is how I solved a research problem in mathematical statistics. The problem concerns robust density estimation a fundamental problem in statistics. Given a contaminated dataset (with Wasserstein-bounded https://t.co/MGgzT4Pfg3 Adding to the list of AI assisting with new math results: Here is how I solved a research problem in mathematical statistics. The problem concerns robust density estimation a fundamental problem in statistics. Given a contaminated dataset (with"  
[X Link](https://x.com/aminkarbasi/status/1998140378328018957)  2025-12-08T21:19Z 11.3K followers, [----] engagements


"@AlexGDimakis At @EPFL_en we used to have oral exams. It was another level to solve multiple (randomly assigned) problems in front of the professor on the spot"  
[X Link](https://x.com/aminkarbasi/status/2001825597471981626)  2025-12-19T01:23Z 11.3K followers, [----] engagements


"I have known @NAChristakis for more than a decade as a mentor and dear close friend. He is one of the most charismatic scholars I know. Do not miss these videos about science. Nicholas Christakis (@NAChristakis) has launched a new YouTube channel For the Love of Science focused on the craft of science how science works and what it is about. Worth a look https://t.co/HlJ3mNkPlc Nicholas Christakis (@NAChristakis) has launched a new YouTube channel For the Love of Science focused on the craft of science how science works and what it is about. Worth a look https://t.co/HlJ3mNkPlc"  
[X Link](https://x.com/aminkarbasi/status/2014012758937817384)  2026-01-21T16:30Z 11.5K followers, [---] engagements


"Hard to have a better deep-dive chat than one with @sanmikoyejo @VirtueAI_co https://www.youtube.com/watchv=JARQT-BicSw https://www.youtube.com/watchv=JARQT-BicSw"  
[X Link](https://x.com/anyuser/status/2016622208702173324)  2026-01-28T21:19Z 11.6K followers, [---] engagements

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