▲ 196 ▼ ChatGPT broke the Turing test — the race is on for new ways to assess AI (www.nature.com) submitted 3 years ago by Five@beehaw.org to c/technology@beehaw.org 129 comments fedilink hide all child comments
[–] Maestro@kbin.social 5 points 3 years ago (1 child) How does ChatGPT do with the Winograd schema? That's a lot harder to fake: https://en.m.wikipedia.org/wiki/Winograd_schema_challenge permalink fedilink source hideshow 2 child comments replies: [–] Droggl@lemmy.sdf.org 2 points 3 years ago (1 child) I dont remember the numbers but iirc it was covered by one of the validation datasets and GPT 4 did quite well on it permalink fedilink source parent hideshow 2 child comments replies: [–] Maestro@kbin.social 2 points 3 years ago* Yeah, but did it do well on the specific examples from the Winograd paper? Because ChatGPT probably just learned those since they are well known and oft repeatef. Or does it do well on brand new sentences made according to the Winograd scheme? permalink fedilink source parent
[–] Droggl@lemmy.sdf.org 2 points 3 years ago (1 child) I dont remember the numbers but iirc it was covered by one of the validation datasets and GPT 4 did quite well on it permalink fedilink source parent hideshow 2 child comments replies: [–] Maestro@kbin.social 2 points 3 years ago* Yeah, but did it do well on the specific examples from the Winograd paper? Because ChatGPT probably just learned those since they are well known and oft repeatef. Or does it do well on brand new sentences made according to the Winograd scheme? permalink fedilink source parent
[–] Maestro@kbin.social 2 points 3 years ago* Yeah, but did it do well on the specific examples from the Winograd paper? Because ChatGPT probably just learned those since they are well known and oft repeatef. Or does it do well on brand new sentences made according to the Winograd scheme? permalink fedilink source parent