-1

AI ethics panels keep dodging the real question about open models

I sat in on a panel last week where three experts could not agree if open source AI should be regulated the same as closed models. The moderator just moved on when someone mentioned that a downloaded model can be fine-tuned for anything without oversight. Has anyone here actually tried to get a straight answer on who is liable when an open model does real harm?

3 comments

3 Comments

the_olivia1mo agoTop Commenter

Totally relatable, it's so frustrating when they just gloss over the hard questions like that.

sanchez.mary1mo ago

You said "gloss over the hard questions" and that's exactly what gets me. Like when I asked my rep about specific policy wording on preexisting conditions and they just gave me some generic answer about "excellent coverage." Did they actually answer your question about the waiting period or did they just talk in circles?

rileygarcia1mo ago

I really gotta push back on this a little. I've actually had the exact OPPOSITE experience with my rep more than once. When I pushed them on the 90 day waiting period for a specific procedure after I switched plans, they actually pulled up the exact page and read me the wording verbatim. It wasn't some smooth talk either. They stuttered a bit and had to check with a supervisor, but they got me the real answer. I think sometimes people go in SO ready for the runaround that they miss when someone actually tries to help. Not saying your rep was good or anything, but maybe it depends a lot on who you get on the phone that day.