Mission: Impossible Languages—What AI Teaches Us About Learning
Can AI learn what humans can’t, or is it just pretending to? A new study challenges everything we thought about language learning.
A team of linguists and computer scientists tested AI language models on “impossible” languages, including grammar no human uses, to see how they learn. Inspired by Noam Chomsky’s claim that AI’s ability to master these languages makes it irrelevant to human linguistics, the study revealed something surprising:
AI struggles with impossible languages, favoring human-like patterns (which makes sense as it is trained on that). Using modified text datasets, researchers trained models to predict word patterns. Models trained on real English excelled, while those tackling shuffled or reversed grammar faltered.
Even a near-English language (“word hop”) caused difficulty, showing AI isn’t the universal learner some feared.
If AI models mirror our language instincts, what does that mean for their role in understanding, or redefining, how we learn?
Read the full article on Quanta Magazine.
----
💡 If you enjoyed this content, be sure to download my new app for a unique experience beyond your traditional newsletter.
This is one of many short posts I share daily on my app, and you can have real-time insights, recommendations and conversations with my digital twin via text, audio or video in 28 languages! Go to my PWA at app.thedigitalspeaker.com and sign up to take our connection to the next level! 🚀