And most (neural network style) "#AI" or #LLM systems cannot even tell you WHY they produced the result they give. It's all in the training data. Huge "garbage in, garbage out" risks/biases!
@i0null@mastodonmigration Counterpoint: human managers aren’t really held responsible for their decisions either, so not convinced that’s a valid blocker anymore
@dogzilla@mastodonmigration if one substitutes the word ‘computer’ with ‘system’ or a ‘collection of prescribed rules’ then semantically the point still applies.
@i0null If everything were self-driving at least we'd know the thread model. It would be like prison sex—you'd know you won't like it, and you'd know you'd be getting a lot of it.
@i0null@publixpace IDK it kinda seems like all the existing self driving cars are super dangerous. Like, way more dangerous than human drivers. And, in theme with this thread, they can’t be held accountable when they run people over.
@i0null the source would be “twitter user bumblebike in 2017”. We only know this slide exists because he shared the picture, and we only think it’s an IBM presentation slide because of his claim.
@i0null I've spent way too long pondering this quote lately.
I think the problem sits in the premise. Is it true that "a computer can never be held accountable"?
And in that, what is "a computer", is it just the physical metal and silicium, the entire product, the entity renting it out, the entity using it, or owning it?
And is it truly "never"? As in: is this something dictated by fysics or the Order of Things, or can this change?
@aiefel@berkes@i0null
And also be careful on pointing to ' creator' instead of 'user'.
Can't blame on car's creator when someone (a driver) kill other with a car.
Can't blame on IA's creator (or whatever 'thinking/intelligence tech') when someone uses it to marking decisions.
@cktodon@aiefel@i0null I can blame the creator if they e.g. forgot to install brakes and that harms someone. Regardless of how stupid I am when operating it, it needs to have some standards before I am to blame.
Add comment