Adversarial Metanoia

Share this post
Human Supervision
adversarialmetanoia.substack.com

Human Supervision

Marcelo Rinesi
Sep 12, 2021
Share

Average government AI engineers removed bugs. The best were tasked with introducing very specific ones.

The desired flaws being impossible to induce without breaking the machine beyond usefulness, engineers (secretly) resorted to manually censoring the results, filtering away their own names, their bosses', and their bosses' bosses' boss from what would have been furiously insistent red flags had the corruption-betrayal detection algorithms been programmed for such emotional nuance.

ShareShare
TopNewCommunity

No posts

Ready for more?

© 2022 Marcelo Rinesi
Privacy ∙ Terms ∙ Collection notice
Publish on Substack Get the app
Substack is the home for great writing