Comment by JeremyNT
11 hours ago
I believe the issue here is that, simply due to how these products came to market, "AI" is extremely vague, and slapping "AI" on every single thing makes it almost a negative signal for quality.
For most users "AI" probably just means "chatbot" - and that's not compelling, because they can already access a chatbot, why would they want one in every product they use?
The more advanced features / workflows that LLMs can enable are kind of opaque if your points of reference are the ChatGPT web interface and summaries of search results on google.com - one reason that "agent" or "harness" have become useful jargon is that it distinguishes the tool we use and what it can do from the tech that backs it.
AI in various forms are used all over, but do your point - users don't know it is AI. They also don't care. They care what AI does, and that is the feature that gets advertised, that AI does it they don't care. They are mostly not chatting with the machines and devices that have AI, they are pushing a button and letting the machine work for them while they sit back and relax (or more often go on to do other things)
There's an almost total, unprecedented disconnect between C-suite perceptions of AI and user perceptions.
In C-suites AI appears to be some kind of limitless source of goodness and profits, so companies must optimise hard for it, or risk getting left behind.
Everyone else is either "Has some uses if you steer it carefully" or "Hell no."
Then maybe we could say that if it's visibly AI, then they've failed. We don't notice the well-done AI, just the badly done ones that hinder the user rather than helping.
And therefore probably in users' minds, when you say "AI", they think of all the badly done ones, not the good ones, because they didn't notice the good ones as AI. So when you advertise it as AI, that's a negative.