5min chapter

Data Skeptic cover image

AI Fails on Theory of Mind Tasks

Data Skeptic

CHAPTER

The Importance of Reproduction in Chat GPT

Professor Kaczynski created 40 vignettes meant to emulate the original tests that were used with children. He took a bunch of variations on the Sally Ann Task and now he couldn't use the original task because these models have eaten up the Internet. The model, not all models, but some of these models say chocolates, right? So you can get them to complete the sentence: In the bag there are, there is, and it will say popcorn. It's smeck-aish-in or something like that. But we change one pixel, and now it thinks it's a bus, right? To me, that's very suggestive about what this thing is doing

00:00

Get the Snipd
podcast app

Unlock the knowledge in podcasts with the podcast player of the future.
App store bannerPlay store banner

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode

Save any
moment

Hear something you like? Tap your headphones to save it with AI-generated key takeaways

Share
& Export

Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode