Brain Snack: Why Humans Learn and Adapt Faster than AI
A new paper outlines what humans do better than AI and how terms can be aligned.
Do humans really learn faster than AI - AI can process so much information?
Yes, AI needs large amounts of information to be able to produce coherent results - or rather the famed Large Language Models such as ChatGPT need, just as they are named, massive databases of language to be able to give coherent answers.
Amazing as this is, and it is amazing, these models also lack a fundamental feature of human cognition according to a review by a panel of 20 cross-disciplinary experts from multiple universities.
And what is that?
It is generalisation - and this includes abstraction.
Us human beings use generalisation and abstraction effectively and can see, for example, an action happening once and from this generalise to a different context. This makes us very efficient and very fast learners.
In contrast AI needs to be trained on a context usually with lots of information.
But can AI also generalize?
There are also models that can do this and this is what this group of researchers also clarified - the different concepts and what they mean in different domains as the paper says:
A crucial yet often overlooked aspect of these interactions is the different ways in which humans and machines generalize. In cognitive science, human generalization commonly involves abstraction and concept learning. By contrast, AI generalization encompasses out-of-domain generalization in machine learning, rule-based reasoning in symbolic AI, and abstraction in neurosymbolic AI.
Are we human beings therefore learning machines?
In some way, yes. I have written previously on how we have a predictive brain and this is part of this ability:
I have also reported on other research analysing how children learn compared to AI with similar conclusions:
But isn’t this generalisation ability also different between human beings?
Yes, actually I also consider it a key differentiator in the business environment. I often speak on the brain related to something in business - some people seem to need very specific examples to be able to apply the knowledge. However, some people seem to have this ability of apply learning to multiple contexts - I see this as a critical ability in the business world and one that is underestimated.
My personal impression is that this ability is getting worse - even before the advent of AI. One of the reasons is we are constantly fed with very specific examples - when I speak at events or conferences I am usually given the mandate to give 3 key concrete takeaways. Which sound like a good thing but it inhibits people making the connections themselves between concepts and across domains.
Can AI catch up here?
Probably - but it may be more complex because we have a physical brain that is structured in a certain way and processes information differently to AI.
I am sure that AI will get closer though. Time will tell how close.
Well, let’s still be happy that we can still do things better than AI, at the moment!
Absolutely!
Reference
Ilievski, F., Hammer, B., van Harmelen, F. et al.
Aligning generalization between humans and machines.
Nat Mach Intell (2025).
https://doi.org/10.1038/s42256-025-01109-4