Archive - Marcus’s Substack

The Fatal Flaws of the Prevailing AI Safety Paradigm

“Interpretability”, “alignment”, and “superalignment” are buzzwords, not feasible safety goals for large language models

Dec 19 •

March 2023

February 2023

Are AI developers playing with fire?

Do developers have any idea what AI are really learning, or how to control them?

Feb 17, 2023 •

This is Marcus’s Substack.

Feb 17, 2023 •

#nojs-banner { position: fixed; bottom: 0; left: 0; padding: 16px 16px 16px 32px; width: 100%; box-sizing: border-box; background: red; color: white; font-family: -apple-system, "Segoe UI", Roboto, Helvetica, Arial, sans-serif, "Apple Color Emoji", "Segoe UI Emoji", "Segoe UI Symbol"; font-size: 13px; line-height: 13px; } #nojs-banner a { color: inherit; text-decoration: underline; } This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts