4 ms·
Not completely unlike with actual humans, based on available evidence, 'talking down to the "AI"' has shown to have a negative impact on performance.
by iugtmkbdfil834 9mo ago
Not completely unlike with actual humans, based on available evidence, 'talking down to the "AI"' has shown to have a negative impact on performance.
- co_king_3 9mo agoThis guy is convinced that LLMs don't work unless you specifically anthropomorphize them. To me, this seems like a dangerous belief to hold.
- iugtmkbdfil834 9mo agoDo I need to believe you are real before I respond? Not automatically. What I am initially engaging is a surface level thought expressed via HN.
- Kim_Bruning 9mo agoThat feels like a somewhat emotional argument, really. Let's strip it down. Within the domain of social interaction, you are committing to making Type II errors (False negatives), and divergent training for the different scenarios. It's a choice! But the price of a false negative (treating a human or sufficeintly advanced agent badly) probably outweighs the cumulative advantages (if any) . Can you say what the advantages might even be? Meanwhile, I think the frugal choice is to have unified training and accept Type I errors instead (False Positives). Now you only need to learn one type of behaviour, and the consequence of making an error is mostly mild embarrassment, if even that.
- co_king_3 9mo agoWhat are you talking about?
- logicprog 9mo ago[flagged]
- Kim_Bruning 9mo agoTL:DR; "you're gonna end up accidentally being mean to real people when you didn't mean to."
- co_king_3 9mo agoI meant to. I want a world in which AI users need to stay in the closet. AI users should fear shame.
- Kim_Bruning 9mo agoReading elsewhere here, you've had some really bad experiences, I think.
- wiseowise 9mo ago> I want a world in which AI users need to stay in the closet. > AI users should fear shame. > created: 3 days ago Unsolicited advice: stop trolling and see a therapist.
- co_king_3 9mo ago[flagged]