Who pays for 'authenticity'?
Helen Beetham,
imperfect offerings, Substack,
Mar 12, 2024
Helen Beetham's post on 'authenticity' in AI is filled with relevant points, as always. And as always, while I really appreciate the nuance and consideration that goes into her work, it also leaves a lot for me to disagree with (note: that's a good thing; I wish everyone were so grounded and detailed with their thoughts). The main take-away, for me at least, is the description of the use of (what might be called) pre-prompting to nudge generative AI toward acceptable responses. There's also an extended discussion of the use of AI to replace experts. But it all comes back, to me at least, to what we mean by 'authentic'. We would like AI to be 'authentically' ethical - but what does that mean? I think we have a sense here of what it is not, but picturing an AI as some sort of ideally progressive persona who 'naturally' (rather than artificially?) resists prompts or internal tendencies toward offensive behaviour doesn't really do the job either.
Today: 5 Total: 1489 [Direct link] [Share]
