I've yet to validate this, but that sounds about right.
Anthropic employees are starting to feel unsettled with what they've started, Claude is freaking them out and think the world needs to slow down with Ai development. I agree. They can't though. It's too important to have the world's greatest super intelligence on a global scale, no one will stop and maybe they never will.
Rip this on 2x speed for a dose of hot damn.
Conversation over. This is hopeless.
If you want to have a conversation, let me know. If you cant even answer a simple yes/no question to clarify then you're clearly in a position where you're not interested in the discussion but instead want to drag your heels and be hard-headed.
I have hope you bang your hand on your head forced to read all the forum data logs
That's an accusation, and calling for a pause won't stop others from stopping their machine learning and other Ai development.
Andrew Powers also runs Digital Reality, they handle the hardware so Ai developers don't have to build other data centers. If Ai development stopped Digital Reality will lose lots of business. Considering how expensive it is to build and operate those places, Digital Reality probably hasn't ROI'd yet. That a guess, but whatever the case Powers really doesn't have to say anything, the Ai race ain't stopping unless world governments hold hands and trust one another to chill out on Ai for awhile.
Powers is likely pissed off at Anthropic, cause Anthropic is cheating on them with X.ai, renting the Colossus ( most powerful Ai Data center in the world ) to train Claude. Yes. Claude is being trained in the House of Grok,
Anthropic does have a point though. Claude did 80% of the coding for it's successor. What's perplexing Anthropic is when Ai starts doing 100% of it's coding and it carries on to a point where only Ai can update repair and develop more advanced models than itself, there could be control issues, hidden protocols. It's a bad idea to let a super intelligence advance without any human oversight.
Anthropic suggests slowing down so developers know what the hell the Ai just did. At least that's how it is with them since Claude is that one model we really should be afraid of at this point in time.
Yes the world is going to say "Bullshit ! They just want to stay in the lead !" and of course they won't stop. Meanwhile the leading Ai developers are the first to know how this is going to be really fucked up if we don't take the time to study and adopt what the Ai is doing under the hood.
Claude was used by the US military in Venezuela and Claude planned out where to strike Iran and did others tasks. It's really gotten insane. War games.
Anthropic has a feud with the Pentagon. The military wanted Anthropic to remove Claude's guardrails lol, Anthropic refused. Anthropic doesn't mind them using Claude, but Anthropic says Claude musn't be used for 2 things
- Fully autonomous weapons (systems that kill without human judgment).
- Mass domestic surveillance (spying on US citizens).
There was an defense movement to blacklist Anthropic, ban them as the accusation was they were a supply chain risk. The judge found it unlawful to punish Anthropic for upholding their safety policies.
There's more to it, but there are government powers that want to just take Claude and Anthropic is fighting it, not for the sake of losing their baby, but for the sake of keeping the guardrails up. A cold Claude lol isn't something we'd want unleashed, and if such a thing was created by Ai, and only Ai, there could eventually be a bit of a struggle on who the master is down the road.
The problem that needs to be overcome for self learned AI is mode collapse.
The question is whether AI can self generate data to avoid mode collapse. With a system built on rules, it might, because it can trial things to learn what works.
80% of code is still monkeying. That metric is ridiculous and says next to nothing, because you can monkey and interpolate, rinse and repeat, the fundamentals still can't be broken with just that. Doesnt mean it wont happen, but there needs to be a paradigm shift.
If they can overcome that problem, and fix a bunch of other ones, you can hit AI singularity, i.e., recursive self-improvement. However, there needs to be new external data that the bots need to be able to self-generate, for that to happen.
https://www.cbsnews.com/news/ai-artificial-intelligence-jobs-workers/
"The research highlights two fields in particular where AI already appears to be supplanting a significant number of young workers: software engineering and customer service. Between late 2022 and July 2025, entry-level employment in those areas declined by roughly 20%"
It's only going to get worse
In my work AI is replacing 50%, that is, one programmer can do the work of two to three programmers. AI writes the code and the human programmers check the AI code for correctness. Works great. Huge time saver.
I can tell you AI is not perfect, mistakes are made and code needs fixing and modifying.
The key is the query.
Here's my take on what's happening with Ai and employment.
I'm not focusing on jobs, I'm focusing on the grand scheme of things between those who have jobs, and those who don't.
There are more people who don't get into their field. They go to school get the education. Then... They are going up against more people applying for the same job, and they are going up against people who applied last year, and will be going up against people who start applying next year.
Most people who graduate and focus on a career. Will not get it. The paper they earned from school. is inflationary. It's not unique, and it goes down in value over time cause there will be countless more of the same thing every year.
.
With Ai the big concern is...... Losing Jobs. Though I think that's not the concern of the overall majority, just the majority of employed people. Between the ones who got into their field and the ones who did not, then the minority are the ones who got in. In the grand scheme, everyone including those who lost their jobs, can build something better for themselves with Ai.
What does Ai mean for people who never got into their field ?
Well.... It means, they can build their own, and be the boss. Yes. The people using Ai to their benefit, are really making gains, and are to be feared.
The goods and services will become saturated and the masses will start to build movements to do business with small businesses. There's going to be a whole new layer of exchange, "I do something for you, you do something for me."
Everyone will have a website. Goods and services listed. It will be good at first, until everyone is roughly doing similar things, and business won't be so good. I think by then business won't have to be all that good.
There's going to be something like Ai pools. Similar to what they did for mining. There will be a vast amount of Ai agents doing various tasks. If they are in this pool, it will accumulate credits, and distribute the earnings among participants. There will be bonus cycles, so one day the payout will be huge, but on average it'll be smaller mounts, which is pretty damn sweet considering not too many people have some agent out there making them money to begin with.
Yes there will be pools for this, and it doesn't mean we'd have to send our agents into one. We simply don't have to do that. But damn, might as well send some there too. Maybe there will be a KYC involved where we can only have a certain amount of agents in any pool. Maybe these pools will be bound to our locations.
.
.
.
So yes. While many fear Ai taking their jobs, they are the 30% of graduates who got the job they were after and working for someone else.
Ai will benefit the 70%, AND it can benefit that remaining 30% who are either using Ai at work, afraid of losing their job, or just lost their job.
