i2km

It's so bizarre. I keep on wondering where the root of this psychosis lies. Is there some common sci-fi literature that sparks these fears among the readers? Or is it all just latent religiosity and spirituality finding an outlet?

I mean, what an appalling way to live. To take themselves so seriously and simultaneously be terrified of what they're building. If we are really seeing the collapse of the bubble now, I wonder what's going to happen to these clowns when their AI god fails... what will they move on to next?

Couldn't anthropic just use fable to find security holes in Alibaba's systems and poison their models?

Or maybe there's been a bit too much hype...

It's somewhat difficult to have any sympathy for Anthropic here. They're entirely responsible for selling tokens at below cost, with the age-old bait-and-switch tactic.

If they weren't doing so, then these Chinese resellers wouldn't be viable. Radical idea, but how about they actually charge a viable price, even on subscription plans?

Surely what's coming is them offering to fix your vulnerabilities via higher-margin professional services?

Shocking. Been open a week and AFAICT just silence from OpenAI. I just find it baffling. You'd think that these vendors would be very sensitive to this sort of issue. I mean, surely they have multiple agents hooked up to github monitoring potential issues and proposing fixes, right? ...right?

Surely it should be trivial for them to have their own tools spinning away trying to fix all the github issues in real time...

Ridiculous. Haven't you heard? All critical thinking skills have long since been sacrificed on the altars of the AI gods and it's inconceivable that we write any code the old way. If you actually understand your code it means you're a luddite and are going to be left behind. /s

My theory is that anthropic have hit the beginnings of model collapse and the whole "fable may silently downgrade with deliberately incorrect results" is a diabolical attempt to gas light and get ahead of the curve.

So when it fails, people will chalk it up to "oh. Must have been silently downgraded because it thought I was doing something tricky enough to count as a distillation attack. My bad. Lemme try again..."

It's bizarre isn't it. It being treated as though actually knowing how to write quality code by hand is a skill that's somehow been lost to time...

I can only guess that a lot of the engineers who went into software purely for money were all too happy to distance themselves from writing code and really don't want to go back...

As a thought experiment, would their reaction have been any different if the hidden prompt had caused their agent to enter an expensive coding loop instead of just deleting the dependency + tests? If I were to use coding agents/LLMs (I don't), this is what I'd be more concerned about...

Much as I dislike some of Peter Thiel's ideas, I fully agree with him that Rene Girard's theory of mimetic desire is at the heart of the explanation. We've always been imitators who need to copy behavior and sentiments, but social media has turbocharged this cycle, allowing mass delusions and panics to be cycled through every few years. The herd always works itself up into a frenzy before collapsing and, a couple of years later, moving on to the next delusion/panic.

And when people are gripped by these crowd-following delusions it definitely leads to the baffling blindness to reality...

If you're willing to go back to the 13th century, Aquinas and the blindness of the mind provide a timeless explanation. They literally cannot see the truth... it's baffling until you realise this and then pity them. They’re in a mini hell of their own making where the LLM is really alive and they’re both awed and terrified of it simultaneously.

Of the same mind. It's a mass panic with overtones of covid. The only explanation I can think of is that most of those who are convinced that AI is going to make engineering redundant are those who've never ventured into the deeper engineering water; never experienced large-scale product engineering and support etc (or have forgotten what it's like).

If there were real gains to be had, we'd see the biggest adopters pulling ahead of their competitors and gaining unassailable leads. It's not happening... Where are the companies which are breaking out and crushing their competitors? Why haven't there been breakouts? Why is software quality seeming (subjectively) to be going in reverse?

All we’ve got is a boatloads of slop, a tsunami of cloned products that their ‘creators’ don’t understand, everything being overrun by bots. It’s just a tidal wave of dross. Where’s the value?

This is one of the things people miss. If they double their customers, of course they double their expenses. Unlike SW, the marginal cost here is still high

The readme is an immediate giveaway of sloppiness

Really appreciate this discussion as I'll be shortly going through this with a 1-2 person company. Does anyone have any experience on how it compares to ISO27001 from the 1-2 person company feasibility standpoint?

Please keep this slop off HN

It’s turning into a bit of a grift now. So many crypto agility “consultants “ popping up with their slop graphics. Never mind the fact that even if a relevant quantum computer is built it will still cost the user millions of dollars to break each RSA key pair…

but the Chinese play long games

And yet they got themselves into a demographic death spiral

Re displacing freelance translation, yes - it can displace the 95% of cases where 95% accuracy is enough. Like you mention though, for diplomatic translations, court proceedings, pacemaker manuals etc you're still going to need a human at least checking every line since the cost of any mistake is so high

This is going to be the concrete block which finally breaks the back of the academic peer review system, i.e. it's going to be a DDoS attack on a system which didn't even handle the load before LLMs.

Maybe we'll need to go back to some sort of proof-of-work system, i.e. only accepting physical mailed copies of manuscripts, possibly hand-written...

LaTeX was one of the last bastions against AI slop. Sadly it's now fallen too. Is there any standardised non-AI disclaimer format which is gaining use?

A blog post looking at developments in the translation market and projecting them onto the future of software engineering. TLDR: software is already bifurcating into low-grade consumer slop where AI lowers expectations, but serious B2B and enterprise software is diverging and (should be) getting better as a result of AI use

One whole technique not mentioned in the paper or comments is bitslicing. For non-branching code (e.g. symmetric ciphers) it's guaranteed constant-time and it would be a remarkable compiler indeed which could introduce optimizations and timing variations to bit-sliced code...

Actually, the link you provide seems to support the parent comment's suggestion, rather than detract from it.

The previous comment was suggesting making sure that every code path takes the same amount of time, not adding a random delay (which doesn't work).

And while I agree that power-analysis attacks etc. are still going to apply, the over-arching context here is just timing-analysis

1984 could only ever have been written by an Englishman

You get the hell out and emigrate. I did so last year. It's not going to get better chap

A quick rule-of-thumb: if 'military-grade encryption' is mentioned, the author likely has no domain knowledge.

Further the article claims that SPNs are used in RSA... which is completely wrong and indicates no domain knowledge.

The article has completely mis-interpreted the paper. The paper is written in Chinese but with an English abstract - the article seems to have just pulled keywords out.

I wonder whether a LLM hallucination is at play somewhere???

The article does not mention AES

So, if the method was to substitute the batteries for ones containing explosives, then how were they triggered simultaneously?

Wouldn't this also require some additional HW/SW in the pagers to trigger the devices? Otherwise, if it was just battery terminals connecting to the battery, how would a remote signal trigger them?

Or maybe it's as simple as the adulterated batteries containing timers and thus not needing external triggering?

Also curious that any mention of confidential computing is avoided throughout the piece

The adoption of advanced cryptography has struggled for decades. It’s a long-running cause of frustration to cryptographers, i.e. “why do we keep on developing more and more advanced schemes yet the real world doesn’t use them?”

This sentiment runs through this piece as an undercurrent.

Part of the frustration comes from the fact, which is emphasised throughout the piece, that all these things are possible in real life. The utopia of autonomous MPC and ZK agents running obfuscated programs to prove identity attributes, vote, etc is all possible, if only we rewrite the entire internet.

The stumbling block, as ever, is human nature. This utopia would decay in just the same way that the bitcoin utopia did, moving from a perfect libertarian idea for an incorruptible currency to a cesspit of fraud, manipulation and speculation.

In Moxie Marlinspike’s article on web 3, he makes the case that the decentralisation promises of web 3 will inevitably decay and be reshaped as a new form of centralisation, leaving us in a potentially worse place than before. Pretty much the same arguments apply here.

Nonetheless, it’s an interesting read. But it would be fascinating to see this approached from a human perspective, I.e. “what are the compelling adoption vectors?” and “how can we avoid re-centralisation?”