How AI Changed the Workplace, and a Lament

It feels like ChatGPT launched only yesterday at the end of 2022, yet nearly four years have already passed.
You can see how quickly AI is changing the workplace. Generative AI is now used across every kind of work.

Planning teams generate planning documents and deliver HTML in place of wireframes.
Development teams write harnesses instead of source code directly, and use AI to generate development documentation.
Marketing teams generate characters and music for short-form videos, then use them as advertising material.

As enormous amounts of money around the world flow into generative AI, the cost of making things like these is falling by the day.
We can make things so conveniently and quickly—so why are well-made products still not appearing?

Because making things well and making lots of things are complete opposites.
The unimportant things made indiscriminately make the important things fade into the background.

A button generated without asking, “Will users need this?” takes attention away from the core features.
Defensive null checks generated by AI swallow exceptions that should have occurred in that scope, causing silent failures.
Low-quality characters for short-form videos and ad creatives can ruin a brand image built with care in an instant.

At work, I see people who have given up thinking for themselves and become slaves to LLMs—so-called Meat Proxies.
People who churn out a lot can look as though they have achieved a lot, and that is likely to feed into their evaluations as well.

We need to think carefully. In this era, a well-made product is the result of people who know how to remove things well.
I was so frustrated by a 40,000-character planning document slapped together with AI that I wrote this post.