- cross-posted to:
- fuck_ai@lemmy.world
- cross-posted to:
- fuck_ai@lemmy.world
This is an uncomfortably well written and thought out article. Its also helped me put into words some of my own conceptual reluctance to fully embrace LLMs in my work. I know how I’ve learned from tackling problems on my own in my career. I worry that the seductive siren of the “right answer” LLM without the effort may be too strong for me to resist or too subtle for me to notice.
Whats the answer? Become the constant contrarian never able to trust anything these output? At that point the value of using the LLM at all is erased.
Bob learned how to use the agent to do work, which is just as useful to society. I wrote a short script to tproxy my connection using an AI and it took less time than doing it myself. More importantly, knowing it took less time gave me the motivation to actually ask the AI to write it
Bob learned how to use the agent to do work, which is just as useful to society.
Bob used the agent to do the work of a junior and it cost him the ability to learn how to do it himself as a senior in the future.
That’s assuming agents can’t do the work of a senior in the future. The future is famously hard to predict
If I am hiring someone, I want to know that when the internet goes down that they can still do something. I want to be able to talk to them about the job without having to wait for them to run my requests through their robot. I want people who can think, not rearrange words based on an algorithm.
And if I was feeling less generous: Bob learned how to ask a robot to do his job? What the fuck do I need Bob for then? Just give me the agent he trained to help him, and he can go find a job somewhere else.
Learning how to use an agent is not “useful to society”. It helps Bob cut corners, but doesn’t extend him. How does society benefit from a summarised email?
Even before AI, when the internet went down I went home. You think I can code without references, stack overflow, etc.? No way, waste of my time
Agree good article


