herknav24

The US needs to confine our military operations to east of 170 degrees west (with a cut out for Alaska) and west of 40 degree west (cut out for Greenland). From Equator to Pole in that range. Everybody else, fend for yourself, we got bills to pay.

@ValerioCapraro It is funny watching people move their AGI goalposts every 6 months or so

Finally, a big name has the courage to tell it: we are nowhere near AGI. Demis Hassabis, CEO of Google DeepMind and Nobel laureate for AlphaFold, put it neat and clear: "Today's systems are nowhere near [AGI]. Doesn't matter how many Erdős problems you solve… I think it's far, far from what a true invention, or someone like Ramanujan, would have been able to do." This is the elephant in the room that many AI enthusiasts prefer not to see, or are actively trying to hide. Erdős problems are well defined, often combinatorial, on finite spaces. They are exactly the kind of problems on which current AI can achieve spectacular performance with a lot of compute and knowledge. A neural network can search a huge graph of possibilities. It can recombine existing knowledge at unprecedented scale. It can discover surprising solutions inside an already defined conceptual space. But true invention is something else. True invention is not only solving a problem. It is inventing new objects, new dimensions, new connections. It is inventing new problems. From resolving to inventing there is a discontinuity that we don't know how to bridge. We are making extraordinary tools. But we are nowhere close to AGI.

@mark_k @giordanorandone My guess is Karpathy joining Anthropic tells us a lot on this. He himself said that the autoreasercher was the final boss battle. I think he wanted to be there for it.

@giordanorandone True, it might have taken a backseat due to agentic coding rush.

So what happened to the Continual Learning breakthrough we were promised in 2026? AI labs have been fairly silent about it lately... Is Continual Learning still going to become a thing or will it remain a pipe dream for AI?

@FLYNAVYBuzzy *Cough* bullsh!7..

@thedogfather @morganlinton Dude don't let a dog barber ruin your day, you doing good work.

It’s kinda weird, people from Cursor, Claude, Droid, Grok Build, and just about every other agentic coding tool seem to interact with me on X. But for some reason, the Codex team doesn’t, and when I comment on their tweets, or tweet to them they always ignore me. So odd because I have publicly shared so many times that Codex is my daily driver and the agentic coding tool I’m the most fanatical about. Moved our whole engineering team off of Claude Code and onto Codex, love it so much. Not quite sure what I did, but would love to interact with people like @thsottiaux, @ajambrosino and @jxnlco - honestly, they’re like my heroes! But for some reason they never interact with me 😢 What did I do guys?

@minchoi @ns123abc He said the final boss battle was autoresearcher for ai/neural net, he joined for end game, he must think anthropic hits it soon.

@ns123abc would've been epic if he joined SpaceXAI

Karpathy added $50 billion to Anthropic’s IPO valuation alone

@nicksortor Do better Nick

WOW! The Pentagon just released a video of an F-16 fighter jet SHOOTING DOWN a UFO Insane footage The second tranche of UFO files is now LIVE at war.gov/ufo

@theinformation Dude nobody gonna do your paywall

Cursor staff meet with xAI employees as layoffs and exits mount.

@nichochar I think he will shift to domain specific models or swing to open source if it looks like anthropic will be first to rsi..and it does look that way.

imo XAI should focus only on this. don’t compete on models nobody doubts that Elon can beat everyone at the data center, factory, space buildout piece

@BenjaminDEKR I considered your view as well. I think everything is a hedge right now, much like the cursor deal. If you are right don't rule out a partnership of more significance with SpaceXai (hardware), Anthropic (models), Mistral (open source).

I predict that xAI (now SpaceXAI) will quickly move away from training their own in-house models, and Grok will stop competing to be a top-ranked LLM. They will focus on being a hardware/ compute provider and leave the competitive models to others (their customers.)

@tulexaicom @jackclarkSF That is correct; as Karpathy said it is the final boss battle. If it happens in 2026 it will be Anthropic, if happens early in 2027 it will be Open AI, later in 2027 my bet is on Elon with the huge training compute advantage.

@jackclarkSF 60% by 2028 from someone with anthropic level access is the kind of probability that should reorganize roadmap conversations. if rsi is real on that timeline, the moat shifts from "who has the best model now" to "who has the best evals when the model writes its own successor."

I've spent the past few weeks reading 100s of public data sources about AI development. I now believe that recursive self-improvement has a 60% chance of happening by the end of 2028. In other words, AI systems might soon be capable of building themselves.

@DavidSacks Mythos slow walk has nothing to do with safety and everything to do with compute limited

It’s time to demystify Mythos. Mythos is not magic. It’s not a doomsday device. It’s the first of many models that can automate cyber tasks (just like coding). OpenAI’s GPT-5.5-cyber can now do the same. And all the frontier models (including those from China) will be there within approximately 6 months. It’s important to recognize that these models do not create vulnerabilities; they discover them. The bugs are already in the code. Using AI to discover and patch them will actually harden these systems. The leap from pre-AI cyber to post-AI cyber means that there will be a big upgrade cycle. After that, however, the market is likely to reach a new equilibrium between AI-powered cyber-offense and AI-powered cyber-defense. Obviously it’s important that cyber defenders get access before cyber attackers. That process is already underway but needs to happen quickly (see point above about Chinese models). Unlike Mythos, GPT-5.5-cyber appears not to be token constrained so it may be the first cyber model that defenders actually get to use.

@mark_k @xai I don't but I was hoping to see some video on it.

@KtdSnav @xai I don't know many who have access to it yet.

Waiting for Grok Build release from @xai. I think it might be a bit delayed because of the trial between Elon Musk and OpenAI. Hopefully this will soon be over and we can go back to business and normal release schedules!

@techdevnotes I'd starting liking your content but if this is how you roll then forget it. Like over other bullshit e-farmer.

Just noticed a reason to have hope in xAI

I'm sorry it hurts your feelings that England has fallen but you should try to take back London from the third world rather than hang on to barren islands half a world away to make you feel like you still have an empire. I'm a restore Britain fan actually but you are all kind of stupid on this one

@KtdSnav @patrickbec @NiohBerg You couldn’t infiltrate a sealed packet of peanuts you daft cunt

Britain could lose the Falklands because of weak leadership. It's not the 80s anymore, there's no Thatcher in charge and the UK is a depressing shell of its former self. If Argentina simply walks in and takes over, possibly soon with Trump's approval, what exactly will Keir Starmer do? Send an angry worded letter? Would he even care at all?

@R_J_Shaw @patrickbec @NiohBerg It is also extremely easy to infiltrate small drone units when your whole defense rests on 1980s warfare. I will be glad to put my combat resume against yours.

@KtdSnav @patrickbec @NiohBerg It’s over 400 miles of open ocean from Argentina to Mt Pleasant. Sit yours ass down.

@patrickbec @R_J_Shaw @NiohBerg 6 FPV drones kill the four hanger queens in the first 8 minutes. Ok now what does UK do?

@R_J_Shaw @NiohBerg Argentina could knock those 4 Typhoons out of action quite easily. A low altitude surprise attack by 8 F-16s should do it. By the time the two QRA Typhoons are ready to start taxiing to take off, the bombs would start dropping on the runways…..

@damonimani Not x users, just engagement farmers

X users: “Engagements are down, my revenue is down, X sucks now!” Me: “I’m glad Elon isn’t in prison, X HQ hasn’t been raided, and I can talk shit on this platform without getting censored or losing my account.”

@OTG_Disciple @CynicalPublius Never will be two guys from Florida

@CynicalPublius I'm not sure Hegseth is a tennable candidate for elected office. DeSantis would be easier to get over the finish line.

I am increasingly concerned with the company Vice President Vance keeps. I will of course vote for him over any Democrat. However. Someone still needs to be nominated. “Rubio/Hegseth ‘28: The FAFO Ticket for America”

@GWOTJoker @CynicalPublius You think the VP is far away from classified information ?

@CynicalPublius TBH if we have to have Hegseth in our government (we don’t) there’s really no better spot for him than VP. Far away from classified information, where he can comb his hair in the mirror all day, and photo op doing CrossFit with Navy SEALs to his heart’s content.