Not really punk but spot-on political: Jesse Welles. The guy is probably the best song writer of a generation.

Also, you asked ELI5:

Because it’s easy to get caught up in the excitement of what looks possible, and it’s easier to appear to be optimistic because of workplace politics. The reality often doesn’t show up until later, when people are already emotionally invested.

You’re not unreasonable. There are lots of very significant problems with AI that make it basically useless, unless prevented. And they can’t all be fully prevented, yet. Maybe never will be.

However, it’s not all terrible to the core.

Full disclosure: I build AI systems for clients. I’m biased.

Most of.my.client projects are spe version of: CEO says “we want AI!” I ask “for what?” CEO gestures vaguely towards the company “for this” We agree on a workflow that a) actually helps them (both the people doing the work and the bottom line) and b) is something where “adding AI” makes sense (without replacing people in the org. At least that’s my personal goal).

Every client project is close to 98% a combination of: I - internal data management II - automation III - tweaking how people are working with tasks IV - getting the people who will do the work excited for the winnings they get and actually change how they work

AI is only roughly the remaining 2%.

Simple things like “Does this user query fit into category a or category b?”

The AI needs a heavily restricted infrastructure to be able to do that properly. The infrastructure and data management and people management are unsexy and most people just skip over it when “building with A”, which is why most outcomes suck. If nothing else because they end up drowning in workslop.

All this to say there is potential for AI to do good things for people, but most people and orgs are just terrible.

Also, all of my clients who, against my recommendation, decided to fire people because of any AI agents or workflows are worse off today than my clients who kept headcount and just made work easier.

Obviously this is assuming things like energy impact, data centres, financials, work safety and accuracy are solved. That’s a very big assumption. I’m working on accuracy.

And, despite working with this, I’m also heavily limiting my own personal AI use and dependence. I totally get where you’re coming from, and I agree with most of it.