Good AI? Part 2, Technology

Restrictions and Acknowledgements Audit and Correct for Biases No More Chatbots Sidebars Direct Validation Limit Scope Session Limits Metrics Next Slide, Please Continuing from where I left off in the introduction post to this little series, I have seen people—whose opinions I respect more often than not—suddenly trying to pitch the idea of “Good AI.” They generally seem to mean something along the lines of exactly what Big AI provides, but trained more carefully and run slightly more sustainably by the masses. And I don’t really get it. Granted, I have spent many hours detailing where LLMs go wrong in my opinion (which still gets occasional updates, by the way) and why I therefore rarely use them at all and certainly never for productive work, but I don’t understand that idea of a good version, except in the abstract. These posts represent me trying to work through the possibilities to find a set of modifications that might result in a system that helps people. Restrictions and…