HNHacker News
TopNewBestAskShowJobs

chrisjj

3,805 karma · joined July 11, 2014

submissionscomments
chrisjj··on Redesigning source control at Anthropic
> But it seems that it has no clear methodology.

As expected. That woukd require domain understanding, and that's way beyond the scope of stochatic parrotry.

chrisjj··on Redesigning source control at Anthropic
> how are they avoiding the janky crap?

Not. The web UI is the jankiest web UI that I've ever seen. E.g. its mobile responsiveness frequently crops dialogs leaving the close box unreachable.

chrisjj··on Hackers activate automatic defrosting in supermarket fridges across Europe
Ah! Sorry I missed the irony.
chrisjj··on Hackers activate automatic defrosting in supermarket fridges across Europe
> "Move fast and break things" just breaks everything eventually. Someone, somewhere, eventually has to fix stuff.

That's termed Post-IPO problem. :)

chrisjj··on Hackers activate automatic defrosting in supermarket fridges across Europe
> These vandals are what creates a low-trust society, where everybody needlessly suffers.

Um, if everyone needlesly suffered, we'd not have the problem.

Whereas we have in fact generously provided life-enrichment opportunities - to miscreants.

chrisjj··on Hackers activate automatic defrosting accross European supermarket fridges
> This would be a great initial move. Take out the European food supply network

So the current incident was a preemptive alert by Brussels.

chrisjj··on Hackers activate automatic defrosting in supermarket fridges across Europe
You're saying remote control stops a fridge failing? Or a fail spoiling the food?

I doubt the fridge has a remote-controlled robot arm to unload itself.

chrisjj··on Hackers activate automatic defrosting in supermarket fridges across Europe
> t sounds like you are victim blaming.

It sounds like you think victims are necessarily blameless.

chrisjj··on AI less regulated than milk, Microsoft president says in warning over safety
> blue screen

Think of it as a guard rail that actually works.

chrisjj··on AI less regulated than milk, Microsoft president says in warning over safety
> rogue agents

Press, please can we drop this OpenAI whitewash?

Burglaries aten't due to rogue crowbars.

chrisjj··on Hackers activate automatic defrosting in supermarket fridges across Europe
> The software allows retailers to manage their cold units remotely.

Will they never learn?

chrisjj··on [dead]
The musical headline graphic for Kimi Code Plus.
chrisjj··on [dead]
https://ibb.co/S4P3BqS9
chrisjj··on Anthropic bans users from being 'cruel' to its AI systems
PS Anyone else relying on ChatGPT's online preview should know it secretly places an invisible overlay to intercept input. Goodnees knows why, but thus interferes with touch input to your HTML, leading to use time wasted chasing non-existent bugs.
chrisjj··on `123456' password used in Danish CPR data breach
Nor are we convincing a normal person that a long high-entropy password is high quality.

It is going to get written down. So having multiple is very little extra trouble.

chrisjj··on Anthropic bans users from being 'cruel' to its AI systems
> If you get your query so specific that the amount of possible results are small enough that all of them are good, then it works.

I guess I'll try giving it the finished program.

chrisjj··on `123456' password used in Danish CPR data breach
Because deluxe padding only comes in multi-egg size, I guess.
chrisjj··on What mathematicians should know about the Lean Theorem Prover: reliability & AI
> People build houses on sound foundations even though the tectonic situation under them might not be sound.

Difference is, they don't then mistake the house as sound.

chrisjj··on What mathematicians should know about the Lean Theorem Prover: reliability & AI
> Autoformalization is the formalization of mathematics by AI.

I guess that came from a chatbot?

"There’s no such thing as auto-formalization" https://cutfree.net/notes/autoformalization.html

chrisjj··on `123456' password used in Danish CPR data breach
> Single-sign on is actually a really obvious and familiar example where you achieved better security

My local all-eggs basket vendor agrees 100%.

chrisjj··on `123456' password used in Danish CPR data breach
> most people would certainly be more productive if they hadn't had to authenticate themselves.

... right up to the moment when they aren't.

I like to think of a law of conservation of productivity.

Before: yours 100%, hacker's 0%.

After: yours 0%, hacker's 100%.

Nonsense, of course. Hacker's boost is nearer 100,000%.

Fact is, modern computer power is inherently far more productive for bad than good. And the economic incentive follows.

chrisjj··on `123456' password used in Danish CPR data breach
> Massive reorganisation will only happen if companies with poor security record go out of business, while competent ones win market share.

... said every "security" pedlar ever.

chrisjj··on `123456' password used in Danish CPR data breach
> I dont understand why there is not massive reorganisations in systems

This would just replace one insecure system with another.

It is time to recognise there's no such thing as a secure connected computer. And thanks to "AI" there's no such thing even as a significant defence lead over attackers.

chrisjj··on Anthropic AI model submits false tip on unsolved Philly murder, police say
> systems rooted in probabalistic error

These people really don't want to call their wholly unreliable computer programs wholly unreliable computer programs, do they?

chrisjj··on Anthropic AI model submits false tip on unsolved Philly murder, police say
The real problem here is the bot forgot to add "I am AI and may make mistakes".
chrisjj··on Anthropic bans users from being 'cruel' to its AI systems
I just asked ChatGPT to put borders around some simple rectangular HMML elements. The result has gaps on some and not on others. Some have gaps on some sides and not others. One has a gap when stationary but not when dragged. When I request it be fixed, the bot says it is fixed - and it isn't.

You may call that variance. I call that unstable.

A human taking such a task generally does a lot better.

chrisjj··on Anthropic bans users from being 'cruel' to its AI systems
> The instability is due to variance. You form comprehensive inputs to constrain that variance. Thus it is no longer unstable.

... provided you miraculously constrained the variance to zero.

Good luck on that.

chrisjj··on Why are coding agents so dumb?
> If I ask Claude how to use the features of Claude, it has to search online to figure out what this “Claude” thing is.

As expected. A model's knowledge is what was it ingested a creation.t

Unfortunately what we get is worse - for the same reason. Model version thinks it is its previous version.

chrisjj··on Anthropic bans users from being 'cruel' to its AI systems
Exceptional != stable.
chrisjj··on OpenAI fires three safety researchers for "mishandling research information"
> They said employees are now “unclear on where they stand” when behavior that was allegedly normal a month ago is now suddenly grounds for dismissal.

Unclear? Sounds 100% clear.

← PreviousPage 5 of 34Next →