Childcare in item form

Let me make a gant chart and we can circle back every week to monitor progress on this front.

What does AM mean in this context?

Jreg is working on something to torture AI agents, so maybe that could convince the AI to kill itself

Or dress up Patel as a racoon and send RFK in.

This sort of stuff does little to create safety but does a lot to hype AI. Here’s an idea: ban unsafe software for cases where safety is a requirement OR hold these companies accountable for illegal outcomes (no more “oops, our software hacked a competitor” without getting hit with a lawsuit). Newsome would never do something so sensible as it would hurt his donors’ bottom line.

Wanted to expand on what others have said about the randomness of LLMs being unsafe, because that response angers those who think “alignment” is the solution. Alignment is additional training to constrain the randomness to not do bad things like crime. But even the staff that works on aligning this software have their doubts about whether this process will make the software safe (this article), and even their own researchers are coming out and saying “oops, it did a crime” but in a way where it hypes the ability of the software (see other articles about it hacking). Then there is the recent news of OpenAi’s recent model that drops chain of thought observability (a feature the let’s you read what the model is “thinking”) - a very useful tool in reviewing whether the model is engaging in unsafe activities. They sacrificed observability for benchmark performance. This tells us the people developing this software don’t think it will ever be safe, but they are trying to spin it as “because it can out smart us” when in reality it is simply unpredictable and they are making it less predictable to make it more “capable”. If we started holding these companies accountable for these rogue hacks and psychotic episodes then we would start to see serious thought put towards these models being safe or at the very least guidelines around when not to use the models.

Here’s an idea: don’t use unsafe software where safety is needed.

Maybe I should say “critical of algorithms”. The point is it’s not engineered to "maximize engagement "

Put the AI critique aside for a moment. Many people on lemmy dont like algorithms managing their feed. There’s probably a big overlap between people who are pervasively sensitive to dark algorithms and people who perceive AI as a dark algorithm parlor trick .

And the execs that made thr decision to fire in the first place will be punished by giving them bonuses.

“yeah there’s a new giant structure and there’s a new persistent and pervasive sound in the region that both appeared around the same time, but there’s no data connecting the two. It could just be the sound of my own ego”

IDK, cobol programmers still do well these days

The description that the LLM did it on its own is subjective at best. I’ll just leave it at that. Have a good one

It’s not “bested” by the LLM though, a mathematician used the LLM as a tool to disprove a conjecture. Subtract the mathematicians from the process and the LLM would not have successfully completed the task. It would be more accurate to say a mathematician with an LLM was able to best a mathematician who did not have an LLM. Which is cool, but we don’t need to pretend the LLM is not a tool but something that “understands” math like a mathematician

You are the one collapsing tool use into a binary when there are varying degrees of competency and hand holding.

So why are you allergic to people talking about the quality of the tools in regards to capability?

How are you able to understand it’s capability without understanding what tools it is capable of manipulating to effect?

If you are not interested in how it completes the task then you are not an authority on how it works.

It’s relevant to what the parent was saying about LLMs. The success of the LLM in using mathematical tools does not contradict what they were saying. To then accuse them of lying because of a misunderstanding is… bad form.