Why Large Language Models give us little insight into language and intelligence


Why Large Language Models give us little insight into language and intelligence — Murat Durmus

Language is fundamentally a social and cultural phenomenon, not a reflection of a pre-existing reality. Therefore, ‘natural’ intelligence requires more than just processing information and following algorithms. Belonging to a language community and having a rich cultural background is essential for language and intelligence.

(I think Ludwig Wittgenstein would agree.)

Everyone knows that culture consists of beliefs, values, customs, behaviors, and artifacts that characterize a group or society. It encompasses all social and intellectual aspects, including language, religion, social organization, technology, and art. Culture can also be understood as the sum of the material and symbolic goods produced by society and passed on from generation to generation. As a result, culture shapes people’s thinking, feeling, and behaving, giving them a sense of identity and belonging. In addition, culture can evolve and change over time in response to historical, social, and political events.

One thing seems inevitable; we’ve already started to create an AI culture, whether consciously or unconsciously.

We live in exciting and, at the same time, challenging times.

So let’s be mindful of this technology and not underestimate the implications for people and society.


Author of the books “MINDFUL AI — Reflections on Artificial Intelligence” and “A Primer to the 42 Most commonly used Machine Learning Algorithms (With Code Samples)



Murat Durmus (CEO @AISOMA_AG)

CEO & Founder @AISOMA_AG | Author | #ArtificialIntelligence | #CEO | #AI | #AIStrategy | #Leadership | #Philosophy | #AIEthics | (views are my own)