The 30% Rule in AI

There’s a great 1983 paper on “the ironies of automation” with interesting thoughts on this. For example, if nobody knows the 80%, how do they gain experience to do the hard 20%? Or, if operators become essentially low skilled, low status babysitters, the smart people required to cover the challenging 20% might just leave for higher status roles. It’s been posted here before, worth a read if you haven’t come across it.

https://ckrybus.com/static/papers/Bainbridge_1983_Automatica...

Stan Sedberry (the author) has had an extremely productive day today. He wrote this article and nine other AI related articles. All today. I am very impressed.

Could have just gone with 80/20 :) Also, AI is definitely more than automation.

The only should is you should succeed in making the change wanted or needed by your customer or stakeholders or yourself. How you do it is a capital problem. Cheaper faster better is preferred.

Imagine saying the 30% rule is you must eat in 30% of the time or eat vegetarian 30% of the time. It’s missing the point.