[temporaryacc2]

The excessive scepticism on Hacker News has poisoned any attempts at rational AI discourse.

The American Government has weaponised state power in a clumsy, corrupt and punitive attack against Anthropic, in an escalating war over control of AI.

Meanwhile, HN has anchored on "marketing hype" as the only possible explanation - all evidence is contorted to fit into this increasingly contrived explanation. Object level analysis is disregarded in favor of dunking on Anthropic.

AI is a threat to your job, status, beliefs, and way of life. For HN, believing this truth is harder than coming up with rationalisations for why it MUST be untrue.

I appreciate the grounded few on HN who continue to engage with object level analysis, and accept that the world is about to change in a pretty bizarre way.

[lubujackson]

This is very reminiscent of the NSA's attempt to limit crypto access of websites in the 90s. We were very close to having the U.S. ban SSL right when it was becoming the clear solution for safe open web transfers for payments.

I assume only the weight of economic pressure and lack of alternatives (and plenty of opposition) kept the doors open, but the same "national interest and security" BS was trotted out then, too.

Big difference this time is nobody needs Fable/Mythos to accomplish anything. There is no magic line here, only improved connective work with less intervention. But it stands to reason that this will cause a huge chilling effect on AI development in the U.S. if it stands and other labs will eventually bypass Fable/Mythos in capability.

To use a car analogy, the models are being built like engine improvements, sometimes going from V6 to V8, but other orgs might improve the car's wind resistance or fuel injection, which leads to a similar speed improvements. There is a lot of space for improvements all along the chain which is why this is such a pointless move.

Knowing this administration (and Anthropic's ham-fisted tactics) this ends in a week or less with some sort of "deal" and was all part of some high stakes negotiation. Possibly even beneficial to Anthropic, because where does it leave OpenAI once they settle on some sweetheart deal? The precedent is set.

[solenoid0937]

Lots of people on HN, and lots of chronic forum users, think that being skeptical/bitter/cynical makes them sound smarter.

Most nerds (like myself) outgrew this edgy mentality in highschool/college. Realistically this mentality just makes it impossible to see anything except through the darkest possible lens.

[boppo1]

This take feels a little like the clergy saying printing presses are dangerous because people will read bad things and spread bad ideas. Turns out they totalky did, but on net it's a small price to pay for widespread literacy.

[temporaryacc2]

If the price to pay is total human disempowerment, I think it's worth getting everyone on the same page before we proceed.

[saurik]

This is a steel man of Anthropic's argument, though, and the premise that there could have been a different thing they claimed that would hold up more doesn't and shouldn't defend their position. To the extent to which it comes down to automating and replacing the need for humans or supporting runaway execution that might accidentally kill all the humans, Anthropic routinely measures it, warns of it, and then releases it. Instead, it is only with respect to specific functionality -- much of which is suspiciously beneficial to them, as they internally claim to use AI to improve their own products while also constantly whining about other people using their AI to improve their products -- that they will put a ton of effort into limiting the access or applicability. The day I boot up Claude and ask it to design a website or automate my paperclip factory and it refuses on ethical grounds is the day Anthropic might seem a little less hypocritical.

[preg_match]

Right, but the widespread literacy took generations to take hold. But, the threat was immediate.

I think, in the long run, of course AI is a boon. But I’m not immortal, and right now it’s a threat to all our livelihoods. We should put ourselves first, and be selfish, while we’re still alive to be selfish.

[rubicon33]

Interesting perspective but don't you think it's too late? Opus 4.8 can already replace most developers. Companies are already cutting the bottom half of the work force, since Opus on a cron can churn out bug fixes.

Maybe there's some hope that if we stop here, we'll still need senior level developers to guide the models and step in where they (occasionally) fall short? Maybe Mythos/Fable was the first time where that was no longer really true?

I'd disagree. I saw plenty of really mediocre work from Fable, especially in various niche programing endeavors (graphics comes to mind). It doesn't strike me as the "end to all programmers" yet... though fast forward 10 years and it seems inevitable that all of our salaries will be halved, if not more.

[paulddraper]

> Meanwhile, HN has anchored on "marketing hype" as the only possible explanation

The Claude.ai web prompt has an undismissable "Claude Fable 5 is current unavailable" notice.

I'm not saying it's 100% a stunt.

But...Anthropic really wants you to know about it.

[solenoid0937]

And if they didn't have this all of HN would rant about how they aren't even telling you that their best model is unavailable, despite having a subscription.

If they tell you, it's a marketing stunt; if they don't, they're being deceptive.

[paulddraper]

I guarantee every frontier lab has an unreleased better model, that they are running through paces.

[simianwords]

I fully agree and this other side of excessive scepticism people are ruining it for everyone else. They are a big distraction. They keep saying things like:

- Anthropic is just doing this for marketing stunt

- AI is like NFT's

- circular deals

- the bubble will burst anytime soon

- the hype bro's are propping up the stock market so that they can exit quick like grifters

(I just made the last one up to force terminology they use)

This is really distracting because the main problem here is that AI is getting too powerful to be just handed out to normal people like us. If you still believe it is all hype, you are getting distracted from the real problem.

I'm guessing at some point this kind of rhetoric will die away and we focus on real problem

[rustcleaner]

>This is really distracting because the main problem here is that AI is getting too powerful to be just handed out to normal people like us.

We need a Second Amendment for AI: the right to keep and bear strong AI shall not be infringed. This safety handwringing is going to solidify the state's monopolies over its subjec... err citizens.

The state is the farmer, and we are the cows.

[rsync]

"We need a Second Amendment for AI: the right to keep and bear strong AI shall not be infringed."

It is already robustly protected by the first amendment.

If I had legal possession of the fable source repository, etc., and printed it on paper, it would be legal for me to possess and distribute, discuss, annotate, etc.

[temporaryacc2]

Safety handwringing?

Mythos found 1000 zero days in a few weeks - if I had asked your thoughts on this a few years ago, I'm sure it would've been "that is a super-weapon".

Plus, scaling laws are impossible to deny: More compute = more intelligence.

AI is going to completely redefine the role of human cognitive ability - if you think this is about "state monopoly", you're really thinking too small.

[rustcleaner]

I want that intelligence in my living room working for me. I do not think Dario, Altman, or the state should get a monopoly on it.

[simianwords]

Ok but you can’t hand wave safety concerns. I agree that they shouldn’t get monopoly over it but what if AI is strong enough to synthesise weapons and help in cyber security?

What’s your answer to it? There are other people who have thought of it and it’s not that simple.

[rustcleaner]

>Ok but you can’t hand wave safety concerns

Sure I can! *waves* Thank goodness we have the First Amendment [here in America] and I can just go to a library to find books with that info anyway.

[DonHopkins]

Yes, you've already told us you're the one who wants to make racist poetry.

[baq]

yeah but they don't want you to and it isn't Dario and sama I've got in mind when I say they.

[rsync]

"Mythos found 1000 zero days in a few weeks - if I had asked your thoughts on this a few years ago, I'm sure it would've been "that is a super-weapon"."

No, my take would have been that a very strong new move had been made in a long-standing, classical arms race.

Since I rather like stable equilibria in areas like this, my thoughts would have been that white hats need to catch up immediately to reestablish the stalemate ... which was probably the course we were on, prior to today.

[boppo1]

Yes, I want that 'super weapon' in everyone's hands. Better than the hands of a few. Same thing as literacy. I believe in the power of the do-gooders to overwhelm the do-badders.

[solenoid0937]

This is the worst possible take. Responsible disclosure shouldn't be a thing? Defenders shouldn't have the chance to frontrun?

[rsync]

I disagree.

See above where I frame this as a classical arms race and the equilibria (stalemate) between white/black hats is a very safe and stable place to be.

You and I don't get to choose who the white hats are and those actors could change dramatically as US administrations change, etc.

It's better to rapidly reestablish a new stalemate ... and I say that as someone who has spent the last six months rapidly triaging vulnerabilities and newly discovered attack surface in an attempt to safeguard the livelihoods of everyone who depends on my business.

[temporaryacc2]

Americans be like

[jbxntuehineoh]

how dare the government infringe my right to possess privately-owned McNukes

[echelon]

Most of the people on HN thinking this stuff is garbage won't be working in tech in five years.

There simply won't be jobs for them.

The risk is that all of these very incredibly smart and disgruntled people decide to do something about it. Elite overproduction, but instead as a result of enormous shift in supply side economics.

[qsera]

Actually I am one of them, and I am thankful for the people who are true believers of AI marketing. Your payments and subscription keeps the LLMS free for people like me who use it as a better search and use it to learn a lot of new things that had no good documentation.

I don't worry about losing my job. I worry about becoming useless. If you know what I am saying..

[temporaryacc2]

If you do not pay for access to the latest models, your experience with AI is a 6-12 month lagging indicator as to current capabilities.

Therefore, it is impossible to have a conversation with you about AI capabilities, because you are anchored on a ceiling that we've long since exceeded.

[boppo1]

I pay for codex & claude. Both out-code me but I'm a novice. Fable is really good and shockingly capable. But they're still dumb as hell in various ways. They're faster than the best humans but they are not better problem solvers, especially for novel stuff like implementing SOTA 3D boolean algorithms in Blender.

[echelon]

For now.

[qsera]

until the money runs out...

[logicchains]

>these very incredibly smart

The incredibly smart ones are able to use AI to multiply their productivity. The ones having a bad time with it from vibe coding and vague prompting aren't that.

[the_gipsy]

Is this "multiplied productivity" in the room with us right now?

[echelon]

I've built one $2.5 million annualized run rate company using Opus this year.

Four months of 50+% MoM growth. I couldn't have done that without the model giving me lots of time to do marketing. And build a complete feature set.

So yeah.

And the year is only halfway done.

[not_a_bot_4sho]

I'm curious. What did you build? Sounds like you might have an interesting "Show HN" post on your hands

[AlexeyBelov]

Nothing. They built nothing.

Proof: I've build a trillion dollar company.

[tripledry]

I'll believe it when I see it.

[lostmsu]

Can't you see there are many people strictly dumber than AI already? And that percentage is rapidly growing?

[tripledry]

> There simply won't be jobs for them.

I simply don't agree with the doomer takes. Might be wrong. I'm kinda stupid yet here I am.

[duskdozer]

True, but not in the way you're thinking.

[sajithdilshan]

This is so true, if anyone posts any positive aspect of AI, those comments are downvoted to abyss. As a software developer I understand how others sees AI as a threat to their job safety and saying AI is evil and must be stopped is so selfish when AI truly can lift millions out of abject poverty in the future.

What’s so funny is that same people are the ones that identify themselves as liberals as long as they can keep their privileged, highly paid jobs.

[latexr]

> As a software developer I understand how others sees AI as a threat to their job safety

This again. For the umpteenth time, not everything is about jobs and money. There are at least a dozen other more valid reasons to be critical or skeptical of AI and the people who control them.

Maybe money and job security is all you think about when you think about AI, but I promise you the rest of the world has many other reasons.

> AI truly can lift millions out of abject poverty in the future.

Pray tell, how exactly will that happen, and what’s the time frame for that future?