> this one is super easy for an MBA to point at and sort of see a way to integrate it
Because a core feature of LLMs is to minimize the distance between {quality answers} and {gibberish that looks correct}.
As a consequence, this maximizes {skill required to distinguish the two}.
Are we then surprised that non-subject matter experts overestimate the output's median usefulness?
Also I think this has been a long time dream of business types. They have always resented domain experts, because they need them for their businesses to be successful. They hate the leverage the domain experts have and they think these LLMs undermine that leverage.
"Business types" get a funny look on their face, when I explain to them that they're the domain expert I seek to eliminate.
In fact, we should try to LLM them away. I wonder, would LLMs then be promoted less?
Actually, I feel like executing this startup and pitching would be hilarious and therapeutic.
"How we will eliminate your job with LLMs, MBA."
I can sort of relate. If you hire an expert, you need to trust them. If you don't like what they say, you're inclined to want a second opinion. Now you need to pay two experts, which is often not reasonable financially, or problematic when it comes to corporate politics. And even if you have two experts, what if they disagree, pay a third?
To manage this well, you need the courage to trust people, as well as the intelligence and patience to question them. Not everybody has that.
But that aside, I think business people generally like having (what they think are) strong experts. It means they can use their people skills and networks to create competitive advantage.
Happens in programming as well, often even by developers.
The "copilot experiences", that finishes the next few lines can be useful and intuitive - an "agent" writing anything more than boilerplate is bound to create more work than it lifted in my experience.
Where I am having a blast with LLMs is learning new programming languages more deeply. I am trying to understand Rust better - and LLMs can produce nice reasoning to whether one should use "Vec<impl XYZ>" or "Vec<Box<dyn XYZ>>". I am sure this trivial for any experienced Rust developer though.