There’s different levels of what is being sold as AI.

At the top, there’s the big promises that AI will replace all workers and somehow usher in a golden age. But there’s never any real evidence to support it. I consider this all hype that’s easily ignored and is only dangerous when powerful people start to believe and act on it.

Below that, AI is coming in like a new tool that speeds up tasks that are otherwise very manual.

Stupid example: remember on The Office when they read Michael’s script, and at one point the “idiot” character was referred to as “Dwigt”? The explanation was that Michael had originally written the character to make fun of Dwight and then changed it via find-and-replace. Except that doesn’t catch Dwigt because it was a typo. An AI solution would be way more likely to catch that and fix it, preventing a humiliating mistake from being printed, and saving a person hours of proofreading.

Better example: when I wrote code at work, I can tell the AI “now write all the automated tests” and in a couple minutes it will spit out a thorough test suite that exercises all the functionality and edge cases. Something that might take a few hours is basically done for me, and I just have to review it. That reviewing would normally have been done by 2 other people before we can ship that change. Instead, I can be one of those reviewers and only one additional person needs to go over it. We’ve just saved somebody an hour.

But those examples aren’t the ones being hyped up so heavily because it’s only making AI “useful”, not “revolutionary”.

It’s a tool. It’s good at some things. It’s hilariously bad at others.

IMO it’s more like how Excel came in and replaced a bunch of tedious paperwork, allowing complicated spreadsheets to suddenly be easily edited and dynamically updated when you need to make changes.