Gabriele Sarti

• Gabriele Sarti 28 Jun 2026 9:03 UTC

  9 points

  0

  on: What did “schem­ing” and “mech in­terp” mean pre-2023?

  Re: mechinterp, I wish everyone entering the field read the great position paper Mechanistic? by Saphra and Wiegreffe, a must-read to understand the field’s origins if you ask me.

A Be­havi­oural and Rep­re­sen­ta­tional Eval­u­a­tion of Goal-di­rect­ed­ness in Lan­guage Model Agents

Model­ling, Mea­sur­ing, and In­ter­ven­ing on Goal-di­rected Be­havi­our in AI Systems