This is a big weakness in FOSS communities, hell, in capitalist existence. People with resources can afford to spend their own time or hire someone else to focus on their contributions like a full time job while most honest contributers will be doing it during their free time because they need to pay bills and such.
Or an alternate implementation of the API that fetches it to flag any programs that call it.
You must be the most dramatic person in the universe, calling that a “meltdown”.
Just flashbang every room before you enter it, just to be safe.
Lol I thought your link was “here’s a rocket designed by an LLM” rather than one designed by the non-LLM AI.
LLMs are a local minimum that tech bros are stuck trying to optimize to a generally useful point because its language abilities are able to fool so many (just like how a real person talking with confidence can fool so many).
This obsession with LLMs is making me question general human intelligence more lol. It’s looking more and more like we are just dumb apes but get lucky and every now and then a smart ape is born and teaches the other dumb apes how to bring their stupidity to whole new levels.
Yeah, it’s good enough that it even had me fooled, despite all my “it just correlates words” comments. It was getting to the desired result, so I was starting to think that the framework around the agentic coding AIs was able to give it enough useful context to make the correlations useful, even if it wasn’t really thinking.
But it’s really just a bunch of duct tape slapped over cracks in a leaky tank they want to put more water in. While it’s impressive how far it has come, the fundamental issues will always be there because it’s still accurate to call LLMs massive text predictors.
The people who believe LLMs have achieved AGI are either just lying to try to prolong the bubble in the hopes of actually getting it to the singularity before it pops or are revealing their own lack of expertise because they either haven’t noticed the fundamental issues or think they are minor things that can be solved because any instance can be patched.
But a) they can only be patched by people who know the correction (so the patches won’t happen in the bleeding edge until humans solve the problem they wanted AI to solve), and b) it will require an infinite number of these patches even to just cover all permutations of everything we do know.
Here’s an example I ran into, since work wants us to use AI to produce work stuff, whatever, they get to deal with the result.
But I had asked it to add some debug code to verify that a process was working by saving the in memory result of that process to a file, so I could ensure the next step was even possible to do based on the output of the first step (because the second step was failing). Get the file output and it looks fine, other than missing some whitespace, but that’s ok.
And then while debugging, it says the issue is the data for step 1 isn’t being passed to the function the calls if all. Wait, how can this be, the file looks fine? Oh when it added the debug code, it added a new code path that just calls the step 1 code (properly). Which does work for verifying step 1 on its own but not for verifying the actual code path.
The code for this task is full of examples like that, almost as if it is intelligent but it’s using the genie model of being helpful where it tries to technically follow directions while subverting expectations anywhere it isn’t specified.
Thinking about my overall task, I’m not sure using AI has saved time. It produces code that looks more like final code, but adds a lot of subtle unexpected issues on the way.
An alternative that will avoid the user agent trick is to curl | cat, which just prints the result of the first command to the console. curl >> filename.sh will write it to a script file that you can review and then mark executable and run if you deem it safe, which is safer than doing a curl | cat followed by a curl | bash (because it’s still possible for the 2nd curl to return a different set of commands).
You can control the user agent with curl and spoof a browser’s user agent for one fetch, then a second fetch using the normal curl user agent and compare the results to detect malicious urls in an automated way.
A command line analyzer tool would be nice for people who aren’t as familiar with the commands (and to defeat obfuscation) and arguments, though I believe the problem is NP, so it won’t likely ever be completely foolproof. Though maybe it can be if it is run in a sandbox to see what it does instead of just analyzed.
I wonder how many billionaires will just get couped by their security forces once society has collapsed to the point where they wouldn’t expect consequences. Their advantages right now come from society as well as the complexity of organizations and social networks. Transitioning to self-sustaining life in bunkers eliminates the power the billionaires hold over the security forces, like what happened with Roman and Ottoman emperors at various points as they grew more dependent on their security to the point the security ended up holding the keys. Especially if they go out of their way up hire security that will do anything they say.
It actually sounds like they are allowing bots as long as a human can intervene and is willing to verify their identity if their automated activity is detected. They are embracing slop users as long as they give them their data.
I think the main reason for the announcement, though, was to deal with the rumours that everyone will need to verify their identity (which this post seems to assume is still the case).
Personally, I’d rather see reddit remain the mainstream version of this just so that people who want to run bots for whatever reason have more reason to do it there than here. Because the commercial level ones will have more resources than lemmy admins will have to deal with it.