F. Abbate, Natural and Artificial Intelligence: A Comparative Analysis of Cognitive Aspects*

Bibliographic information

Abbate, F. Natural and Artificial Intelligence: A Comparative Analysis of Cognitive
Aspects.Minds & Machines 33, 791–815 (2023). https://doi.org/10.1007/s11023-023-09646-

Commentary

The author has a pleasant way of writing and organizes the paper in a logical way. I think
there is, however, an excessive dependence on philosophical frameworks. As a result,
its conclusions are less applicable in practical settings. It might even result in circular
reasoning, where ChatGPT is considered non-intelligent since it does not fit the
definitions of intelligence that are centered around humans.

“I define intelligence as the ability to adapt to the surrounding environment and deal effectively with new situations. Adaptivity intrinsically implies […] relevant intentional agency. ChatGPT’s behavior is relevant but not intentional.”

This is a logical definition of intelligence. I tend to agree that ChatGPT’s behavior is not intentional. I do question what exactly is meant by ‘surrounding environment’ . Could the coding architecture and the technology on which ChatGPT runs be seen as its environment?

Well, ChatGPT has been shown to be capable of a range of tasks even if it was not
specifically trained for them. The paper states that it can ‘solve difficult tasks that span
mathematics, coding, vision, medicine, law, psychology, and more, without needing any
special prompting’. To me, this shows that even though ChatGPT was given a certain
task (to predict the next word token), its skills were transferable to all these tasks. So,
setting up a certain environment does not necessarily have to mean that it is limited.

“The learning realized by the program contributes to an intensification of these faculties, without the program being able to dispose of them in the proper sense being it not self-aware.”

So, in other words, for something to be intelligent, it must have some self-awareness in
order to activate certain faculties with intention. The author claims that his definition of
intelligence applies to all levels of cognition. However, he also ascribes the minimum
level of cognition to Escherica coli, a bacterium, where ‘cognitive faculties emerge
evolutionarily, in response to environmental problems, and are characterized by the
success of the action’. In my opinion, this statement does not include any self-
awareness. The bacterium does not consciously choose to do certain actions. So, the
link the author sketches between intelligence and cognition and between bacterium
and ChatGPT is not adequate.