ACB

@hughesanalytics @skirano Ffs…. Just when it starts working 🤣

@skirano That I believe!

Normally I fight back against posts like this but I have absolute respect for Pietro. Someone is going to have to explain this to me like I am five why this is true.

@hughesanalytics 🥲

@ACBPhysio I spoke with my sister as she is in the UK. Her point of view as a life long conservative. Reform would win a fair general election. Labour and the Conservative parties will do everything to avoid a fair election or any election at all. That is the UK at the moment.

I can't believe it but here we are.

@emollick At least the US has some frontier models…. Unfortunately, the EU lagged totally in this field… it’s ‘golden boy’ Mistral doesn’t make up for it.

It is possible that xAI catches up, certainly Musk has stated that he thinks they can in a year. And Meta still has excellent people and lots of compute. But if you view the critical period, for better or worse, as this year, it is a bad time for a lab to sit this out right now.

The failures of both Meta and xAI to maintain parity with the frontier labs, along with the fact that the Chinese open weights models continue to lag by months, means that recursive AI self-improvement, if it happens, will likely be by a model from Google, OpenAI and/or Anthropic

@emollick Congrats, Professor! Ever since the book and the substack learning a lot

Two and a half years after we released our paper (which both coined the phrase “jagged frontier” and provided some of the first experimental evidence of real productivity gains from AI), it has now been published. The academic process takes awhile! Read: pubsonline.informs.org/doi/10.1287/or…

@hughesanalytics If I’m expressing this shared feeling to others…. The don’t seem to understand, they think it just makes you dumb 🤷🏾‍♂️

@ACBPhysio It really is. I am quite conflicted. I could see myself devoting 100% of my time to this.

Switched to GPT-5.4. It feels so different from GPT-5.2. Very impressive so far although I didn't really feel I was missing anything with 5.2. Certainly more analytical with a different voice.

@hughesanalytics What a time to be alive… 😅

@ACBPhysio These models are getting so good. If there was no improvement from here on I would be fine. It is incredible.

@hughesanalytics Indeed… big Gemini Pro 3.1 difference and functionality. Got me like:

GPT-5.4 is freaking me out a bit. It is shockingly good. Feels alien.

@KatanaLarp @choffstein Nice! First things first; set up the damper and drag factor. m.youtube.com/watch?v=piJD5b…

got a rower now. who am I up against? @choffstein ?

@GregLehman Grifters gonna grift

Let me translate what this dude says: "Wishing her the best now let me tell you why it's her fault. Hey, who wants to buy some BS injections?"

@tradewarfan @mikeharrisNY Agree on the math and critical thinking (although the concept of critical thinking is still ill-defined). And how does one acquire or train “adaptability skill”?

@ACBPhysio @mikeharrisNY For me, it's solid math, critical thinking, and adaptability-skills that help you thrive no matter how fast tech and the global market change.

One problem is frequency of the change in suggestions is higher than the time it takes to get a degree. Those same people were told 4 years ago to learn how to code. As soon as they coded LLMs, now they must learn physics. Then what? LLM learns physics? Technobros.

@tradewarfan @mikeharrisNY Please elaborate what the ‘solid fundamentals’ are in your opinion?

@mikeharrisNY I swear, by the time you graduate, the advice resets again. I stick to building solid fundamentals-way more reliable than chasing the latest tech trend.

@mikeharrisNY I saw the open and thought

I have great respect for the counterparty that just paid 3 months worth of lunch at Five Guys. And BTW, "I told you so". 😂 $QQQ

Is there anything in the local LLM sphere as stable and fast as IBM Granite 4H Tiny? Damn, impressive. We should push further in this direction (low energy, high accuracy) in development.

@hughesanalytics Just say when and I’m subscribing ASAP. I’d be reading out of interest and to learn your views on AI/ML topics. I wouldn’t mind the occasional financial chart or 💎.

Lastly, X is probably not the best place to talk about this. I can imaging that it would be incredibly annoying to see charts of different contracts and timeframes. Substack is an option. DM for details. There might be some AI/ML discussion as well.

@lukeburgis We need some guidance on reaching a ‘writer 4’, because there are so many deceptive moments when using AI, where you think you’ve achieved a deeper moment or experience but actually just get more brain rot. What are the boundaries, where to summarize and where not?

Quite a few people have been surprised at my Clawdbot adoption and usage. Yes, I have it set-up on a dedicated Mac Mini in my office—and not because I'm quite THAT mimetic, but because this is what an extremely trusted former head of security for a hedge fund advised me to do because it was the safest option; the AI is totally cordoned off from most of my life from a security perspective, and access is granted very intentionally. Let me explain a bit my reasoning why I, at least in my capacity as an author, think that double down on personalized, agentic AI's is a smart bet (and upon doing enough research and talking to enough people, I recognized Clawdbot was the real deal—and the pat 4-5 days have proven to be the most exciting time I've had with tech since I was 12 years old and built my first computer): Short pre-amble: I want to be more creative. I want more time spent on cultivating and communicating original ideas, connecting them to others, and communicating them to the world. I also have a growing operational footprint that is massive: between directing a growing institute, being a university professor, the role of investor, a board member, and more, not to mention a father of two under two and a half, I have more operational and logistical challenges than I've ever had up until this point, like running 3-4 companies at the same time (I am). My inbound communication alone is impossible to keep up with, even with an EA, and things are only getting more complex. The world is getting more complex. I'm heavily utilizing AI on the operations side of my business and platform—not the writing—to help me do more of what I love, to stay focused, and to keep up with responsibilities. Looking back in 10 years, I think we will easily see that there were four kinds of writers: 1) The ones who disappeared from getting caught up in the slop—who use AI to the point that it ruins their writing, their minds, their souls, and who disappear into oblivion as undifferentiated, derivative of the machine. 2) The ones who shun tech and AI altogether and may or may not produce good writing—but for the sake of argument, let's say that their decision actually allows them to improve their craft and avoid the pitfalls that AI can introduce if not carefully managed; let's assume they produce great work, some of which is recognized immediately and other work that isn't recognized until 50 or 100 years from now. I respect them—it's just not the route I'm taking, and I suspect their decision will actually leave a lot of serious human value creation on the table due to the people who will never know about their writing or ideas for years, maybe even decades. 3) The writers who prudently use AI as an aid and supplement for research and possibly even knocking around writing ideas, but who don't truly operationalize it. Their writing has some rounded edges due to the sub-optimal AI use giving them AI-brain; they might post a banger tweet from time to time, or produce an interesting essay, but they are essentially floundering, keeping heads above water, using AI as a crutch, tinkering here and there but only taking advantage of 1% of its true capabilities—which ultimately means they get few of the real "deep" benefits, but a lot of anxiety around navigating "how to use AI so as not to get left behind" etc. Don't do this. 4) Others will actually use it to extend their writing into domains they couldn't have even imagined just a few years ago. They produce excellent raw material, they are the talent—and the AI operationalizes a system that allows them to do more of what they do best, to focus, to write, to think, rather than the administrative work that holds so many of us back. They will use technology in such a way that it leads to an opening of the spirit, not the closing. I intend to be in the fourth group. 5 or 10 years from now, we'll look back and realize that these were the lanes people were choosing to run in without necessarily knowing it. I'm looking for others to run alongside me. I plan to be writing (longer form) about these topics in the coming days/months, now that I've just submitted a manuscript that I've spent the better part of the past two years writing. I'm willing to build in public. And I think those of us who are both serious about the craft, and serious about the future, should get even more serious.

@hughesanalytics How about medical use cases? Private practices wanting to use LLMs with greater privacy protection or their own data?

I have these in a different format and they are really only suitable for running small models. Once you are past 12B parameters the response time just isn't good enough. They are good general purpose computers.

@windscribecom @zaingardezi01 @Cartidise Where to self host?

@zaingardezi01 @Cartidise if you don't wanna self-host, Proton or Tuta are both fine for most people.

Proton, you ok?

@kareem_carr Math or Carpentry’s?

I'm increasingly thinking the best thing one can do right now is just skill up. Read some books. Learn some math. This politics is unstable. It won't last. When it's all over, we'll need people who can build.

"What color is your Bugatti?" -Andrew Tate Don't be suckered into the cult of "Winner Culture." No one is going to tell you this, but if you just take life advice from Pro-Athletes, Billionaire Moguls, VCs and other 'Winners', you are not going to get the full experience. Consider the amount of wisdom held by people who will have well-attended funerals, yet who never 'fly private'.