If you can’t make things right when your agent goes wrong, then yeah you’re not adding value. If you can’t tell the difference between the agent hallucinations and the useful stuff, then yeah you’re not adding value. If you can’t steer the agent in a useful direction instead of just standing back, then yeah…

▲ _diyar an hour ago | parent [-]

You sound dismissive, but taking your description to its conclusion, the engineer value can be approximated as:

  {(P(AI mistake) * P(human spots the mistake | AI mistake) * P(human fixes mistake | given AI mistake, human spots it) - P(AI mistake)} * cost of AI mistake

For many engineering jobs, P(AI mistake) is very low and decreasing, and cost of AI mistakes is also usually (consider the cost of a bug for an average engineer; unless you’re building avionics or working on MRIs or such, usually very low).

So even if you can “make things right” or “tell the difference” as you say, the value of any given engineer is dropping.

▲

bryanrasmussen an hour ago | parent | next [-]

it's worse than that, as you need to also compare

human mistake, AI spots the mistake

Human mistake, human spots mistake

AI mistake, AI spots the mistake.

Everybody is always talking as if they don't make mistakes, but every software engineer does; it used to be that human caught those mistakes and fixed them, but if AI catches those mistakes and fixes them, then that is also a decrease in value of human.

	▲	_diyar 11 minutes ago \| parent [-]
		Human - Human and AI - AI mistake and spotting thereof basically cancel each other out, or can be considered no mistake. Human - AI just speaks in favor of AI, and my comment was responding to somebody dismissive of how crushed engineers can feel by AI.

▲

bigstrat2003 an hour ago | parent | prev [-]

> For many engineering jobs, P(AI mistake) is very low

Not remotely the case. LLMs make mistakes very frequently, much more frequently than humans do.

	▲	bryanrasmussen 44 minutes ago \| parent [-]
		source? I'm wondering at the curve of those mistakes actually, I doubt modern LLMs make mistakes at the level of the worst engineers I've worked with. My last project had 2 guys from India and one guy from a consulting company that I take any LLM over them. The guy from the consulting company spent more than a month trying to implement a design that actually was not implementable. I obviously also blame the supposed professional design company that produced the design, and I mean it took me about 6 hours spread over a day and a half to figure out waitaminute this is actually not doable! But I have to wonder how badly would an LLM have done. I sort of think the same, because probably an LLM is never going to announce - hey, this design as stated is impossible, some parameter needs to be changed.