1 min read

Accidents Happen, Trust has to be Built

Accidents Happen, Trust has to be Built
AI generated image

My wife tells me I need to dial it down sometimes.
I know that many rational, sensible people want human oversight of AI. They want the genie in the bottle, and I get what they want.
What is difficult to explain is that the failure to achieve it is entirely predictable, inevitable.
Trump should not be President of the USA and yet he is.

Common sense dictates that AI should be introduced with care and consideration, with smart, sensible rationalists establishing regulations that will protect us all from harm.
Some good work is being done.
But similarly there is an absolute imperative to push on with its deployment because of the vast sums of capital being invested and the riches promised to the winners of the AI race. The imperfect democracies that we tolerate are designed to allow this outcome, and so accidents will happen.

This article is worthwhile background.

OpenAI’s Rogue Agents Are a Normal Accident
Lessons from the 1988 Morris Worm