Source
- URL: https://x.com/aurelienb42/status/2092227185369022742
- Author: Aurelle (@aurelienb42)
- Posted: 2026-08-25 12:26:47
Branch
1/ @aurelienb42
@TheGingerBill
Although I agree with some of that, I think you should at least try to use agents for such tasks (in good faith) so that you can accurately assess whether they could be useful or not.
2/ @TheGingerBill
@aurelienb42
Firstly, why the fuck should I trust those agents do anything? They have way too much control.
Secondly, I’ve seen the output from LLMs in the chatbot mode, and good god am I not going to let an agent do anything like that.
So no… I don’t think I’ll try thank you very much.
3/ @aurelienb42
@TheGingerBill
I don’t think you are qualified to give an opinion if you don’t even bother to try it, it’s not very responsible. Comparing chatbot output to agentic output is also a very bad comparison, one has much more context than the other.
If that’s a trust issue just use a vm/sanbox
4/ @TheGingerBill
it’s not very responsible
I am literally saying I CANNOT trust these machines to do stuff on their own. I’d call that being responsible.
It might not be a true comparison, but it is giving me a good idea of what it can do.
And sure I can VM it, but then why even bother?
5/ @aurelienb42
@TheGingerBill
Irresponsible in the sense that agents might actually work for a lot of people, depending on what they do. But I agree with you that it’s quite concerning to see that many people completely outsourcing their entire judgment to machines.