knowledge cutoff staying the same likely means they didn't do a new pre-train. We already knew there were plans from deepmind to integrate new RL changes in the post training of the weights. https://x.com/ankesh_anand/status/2002017859443233017

This keeps getting repeated for all kinds of model releases, but isn’t necessarily true. It’s possible to make all kinds of changes without updating the pretraining data set. You can’t judge a model’s newness based on what it knows about.

I suspect this may be the case. There’s inherent inefficiency in having a human forced to translate everything into context for the LLM. You don’t get the full benefit until you allow it to be fully plugged in.

This is the biggest bottleneck. To realize the “replacement of white collar workers” fever dream, (which is, I still believe, technically feasible), you need the agent that replaces them to have all of the context they had. All of the emails, all of the Slacks, all of the meeting minutes, access to private corporate systems and files, etc. I can’t think of a single company that would want to turn all of that over to OpenAI.

> I can’t think of a single company that would want to turn all of that over to OpenAI

you’d be surprised… the largest IP in the majority of cases is the codebase itself. once that hurdle was crossed the rest is easy decision

Interesting idea. I do feel one of the major barriers to mass replacement of white collar workers is lack of direct integration between email/slack and LLMs. A human still needs to distill organizational needs from multiple stakeholders to write a prompt.

At the same time, I would be very surprised if companies are lining up to hand over all of their internal comms to OpenAI. Would need really strong privacy guarantees, and I’m not sure OpenAI has the goodwill to be convincing on that front.

Also, doesn’t MS still nominally have some stake in OpenAI? Would be surprised if they were chill with another competitor to Teams getting built.

Openclaw has that ability. Read emails, take actions.

Also, from the sounds of it the Ms OAI deal is in its last days.

Fiverr CEO goes on a pretty dark rant about how people who don’t upskill and convert to AI workflows will have to change professions, etc.

Then concludes his email with:

> I have asked Shelly to free up time on my calendar next week so people can have conversations with me about our future.

I assume Shelly is an AI, and not human headcount the CEO is wasting on menial admin tasks??

I assume Shelly Paran is Fiverr's Chief of Staff, and a human. You already knew that rules don't apply to CEOs.

The irony is if you take this to the limit, ChatGPT replaces fiverr

China innovates orders of magnitude faster than they did even ten years ago. Yes, a lot of it is still copycat, but there is value in being able to copy quickly and well.

The question is whether China offers long-term stability for external investment. Should US retirement portfolios load up on Chinese equities?

> So, perhaps we won? Perhaps we built our markets so stable that they are these days impervious? That sounds silly on its face, and the two reasons I’d actually give are:

> 1. Markets are just slower moving than ever before, big players just like to sit on their big piles of money

> 2. There are one or more bubbles in the stock market. Almost everyone agrees that AI is a bubble. It funds itself in a circular fashion, and capex cannot be recovered with profits any time soon, even with optimistic outlooks.

It’s a bit of both. The impact of political instability in the US (read: Trump pissing off as many people as possible) may not be felt in the markets quickly, if even within his term. He has severely dented confidence in the US as a trading partner and as an arbiter of the global rules-based order. That will have decades-long implications, the result being a pivot away from dollar-denominated commodities trading, and export markets for US goods being increasingly unfriendly. The value of the dollar will probably decline, and in fact that is a goal of many in his administration. That could actually be good for US equities if it’s in moderation.

The biggest risk I see is flight of capital away from US treasuries, which would drive up interest rates, leading to a sovereign debt crisis in the US. The likely solution to that would be money printing and resultant inflation. The high treasuries rate would drag capital away from equities.

Exactly. Trump is just the messenger. The underlying problem, festering for decades, is that a large enough fraction of the US population thought that Trump was even remotely qualified to lead a country to begin with.

The fact that we didn’t immediately remove him after the Greenland letter just demonstrated to our allies that that same fraction of people also don’t value our relationship with Europe.

The CCP looks thoughtful and stable by comparison. It’s too bad that the result of this will be to force more people to align with them, because they’re just as self-serving as Trump, they just have the good sense and temperament to hide their true intentions.

The fact that we didn't jail/banish him after leading an insurrection against the US Capitol was the "too far gone" moment.

Anything after that, is just Yet Another millimeter along the same trajectory:

Well, if we let him get away with J6, then I don't see how this is too much worse.

To tie this to your point, I honestly don't see why the Greenland letter should suddenly be the straw that shatters this particular dromedary's spine.

Nicer to ride in in what sense? I find single-pedal to be nauseating because many drivers can’t control their foot raise well enough for smooth or gentle braking, and the suspension feels chunky as hell, most likely due to how heavy the car is. But maybe that’s just Teslas. Their ride is just categorically worse than most ICE vehicles.

I like this viewpoint - it basically casts VC-backed AI startups as privately-subsidized applied R&D projects, which largely seems to be the case for foundational model companies.