> If Claude gives me poor or incorrect advice while I’m working on an AI component, I have no way of knowing whether the model was confused, whether my problem is unsolvable, or if some invisible policy restriction quietly kicked in.
You should be able to know if your problem was solvable by using your own expertise and judgement, no? If you're relying on LLMs as a substitute for those, I wouldn't expect great results.
No; once the LLM switches to this new saboteur mode, it’ll be very hard to detect.
Sabotage is an asymmetric weapon. The ratio of damage to effort is nearly unbounded, and any decent saboteur knows that the key trick is to make your output indistinguishable from incompetence.
They’re building state of the art offensive capabilities into a public model, then expecting to maintain control over when it decides to attack its human users.
The premise is laughable, and we’ve all seen how this movie ends.
I honestly feel like that's a counter-example. With AI he'd be tweeting some other nonsense. It's not like anyone saw the image and thought he actually was orange Jesus.
Multiple people have the same response, I randomly selected this one for follow up:
Yes, but two things were lost:
1) the need for skill or an accomplice. He _couldn’t_ tweet that image in 2016, not without first asking someone to photoshop it. And that need to engage in human to human communication is something truly fundamental that was changed and lost.
2) Any ambiguity or misunderstanding. Yes bad textual tweets exist for a long time in politics. But there IS something about images that is more powerful than text. The text „I’m Jesus Christ and god sent me to heal the sick“ would probably make the news, but a lot of people would go: „is he quoting the bible? What’s going on?“, not so much with Jesus Picture.
I'd hoped this would be a post about Giro Galoppo, which is a fantastic little board game if you have kids (and can find a copy). It's easy enough for small children to learn, but the mix of strategy and luck is such that an adult can put real effort into winning, but will still probably only win 2/3 to 3/4 of games, not every time. And it's quick to play. Great game.
Certainly, and some discoveries have been attributed to AI (I was reading that mozilla firefox were praising mythos recently)
But that's not accounting for all of the discoveries, not at all.
I've also seen the npm people talking about the surge in AI code overwhelming the ability to properly review what's being distributed, and a large number of vulnerabilities being attributed to that
Very much another "Emperor's New Clothes" situation.
If the pathology was entirely within his own privately-owned company that'd be one thing, but Americans are going to continue to get hurt because of it.
How is it an anomaly? Global stocks go up (or at least have positive total return) over time on average because companies produce value. Ultimately it's true that that money has to come from somewhere, so you can say printing it is monetary policy, but the reason it can be done without runaway inflation is the tangible value produced by the firms.
Do you think carrying a gun gives the child soldier power to change his circumstances? It's a problem of organization and collective action, not lack of arms.
You should be able to know if your problem was solvable by using your own expertise and judgement, no? If you're relying on LLMs as a substitute for those, I wouldn't expect great results.
reply