[RugnirViking]

does it never? seems to me that people pay me precisely for my knowledge, learned over many years. The knowledge translates into action, sure. But thats like the old parable about a plumber being paid €150 for a 5 minute consult that involves turning a single screw. "i could have turned that screw!" the customer cries, ignoring that yes, they could have. But they didn't know to.

I think perhaps the problem is instead "I thought that having this knowledge would set me apart, forever, without me having to learn anything else"

[esikich]

There's a good chance the apprentice plumber could've fixed it just as quickly. That's where we are now.

[RugnirViking]

right. Apprentices will always grow, and so too must you, if you want to keep being paid. Their job is to come with new tools and new ideas, and your job is to keep a wider view into what you're doing and why, maintaining trust (you need to build the authority to tell apprentices no when their ideas might flood the customer's house), and keep moving towards other parts of the business and solving harder problems (working with sales, hiring, etc to manage customers and apprentices). AI will not build authority for you.

If your argument is that the customer themselves could use an AI or whatever to learn plumbing, that was always an option (libraries, google, youtube). They pay you so they don't have to worry about flooding their house (or at least have someone else to blame).

They might be able to "one shot" simple fixes that you might previously have assigned to an apprentice, but believe me, AIs are not about to start doing complex things for the layman that actually required seniors previously in either programming or plumbing, because very few of those things were just "type better into a computer". (build trust, speak confidently, know what doesn't work, take responsibility, test without breaking systems, communicate and work together with other professionals, have opinions)

[ufocia]

Libraries, Google and YouTube were/are not nearly as efficient at conveying _targetted_ _actionable_ expertise as AI is.

[RugnirViking]

I agree that it is easier than ever to start doing stuff, instead of reading. I don't think that means its easier to jump right to doing large projects. The problems to be solved there are often subtler, of a different class, and manifold, and a layman may not realise what has gone wrong until long afterwards or never (this also happened before, many people took on projects they weren't ready for and reinvented the wheel trying to solve issues they ran into)

it's oft debated, but I do fall on the side of "you should still know maths even in the age of the calculator/matlab/llms". I have found productive employment, and indeed tickets to speak to the big boys in their gilded palaces many times because graphs and charts are their favorite toys and knowing maths got me there. They have always been able to make things with excel, with matlab etc. Often they actually can make charts themselves, but they don't care to become experts in what data is important and what isn't.

The LLM isn't yet good enough to tell you what data matters. People act like LLMs are magical gods that do everything, but it is but another tool. It has limitations, just as it has strengths. It is not ultimately convincing, it is not infallible, and experts will keep finding edge cases all the damn time. Anyone working with them every day knows this, and you need to know it too.

[ValentineC]

On the flip side: it's trivial to search "how to fix that pipe" on YouTube, see a bunch of success videos, and trust them all.

I'm not sure I can trust any single AI, or even multiple AI models, to not hallucinate overconfidence in certain real world domains.

[smcg]

targeted, expertise, fast... pick 2

[altmanaltman]

I think a more sane minded customer would not mind paying for the assurance and having someone to blame in case things go wrong, not necessarily because of their domain knowledge.

I could theoretically learn everything about plumbing but would still rather call a professional for the peace of mind that it was done "correctly" and it the process goes wrong, I would have an instant fix instead of trying to go back and educating myself on plumbing more.

Could you consider that as part of knowledge? Yeah and also no. Because the knowledge can be copied and put into a LLM but legally a LLM cannot sign off on things like NDAs or take accountability like a human has to in these roles.

[RugnirViking]

I agree. I also think that deciding that LLMs encode all knowledge perfectly, either now or in an imagined future, is foolish. My experience is that they match the average general state of experts among the field. The sort of thing a junior might read to start to grasp the general ideas and issues in a field. They rarely have opinions, or good intuitions around more specific scenarios. This is why the current equilibrium of a senior piloting one works so well- theyre leaning on it to speed up, but pushing it away from the "average" where circumstances demand.

We can argue about imagined future progress, but I don't see that getting much better, given that the literature doesn't often do that, and how often experts in one scenario end up being poorly suited given another set of facts.