The embodied, interactive, multi-modal nature of human first-language acquisition is hard to replicate in current AI models.

Added to korrents 20 Sept 2026 · How quotes work · Something wrong? Tell us

Shared subject first, then similar wording. They may agree or disagree.

• The hope for AI interfaces is that people will not have to pick a modality up front, and can move fluidly between chat, point-and-click and voice. CRCarl Rivera10 Nov 2025
• He reached the conclusion that input drives language acquisition on his own, before meeting the research that says so. SKSteve Kaufmannby 20 Jul 2026
• Humans can generally perform new language tasks from few examples or simple instructions, while current NLP systems largely cannot. DADario Amodei28 May 2020
• Duolingo's real competence is motivation and habit formation, and language learning was only the first thing it applied that to. ACAlbert Cheng5 Oct 2025
• Current techniques restrict pre-trained representation power because standard language models are unidirectional. JDJacob Devlin11 Oct 2018

See this claim on the map