Here’s an attempt to describe human intelligence using terms that we use to describe artificial intelligence:
Human intelligence is characterized by his capability to act independently in terms of setting and updating his own goals, planning how to achieve them based on available resources, and adapting timely to arising changes in the surrounding environment. These capabilities are associated with other human cognitive abilities like being aware or conscious of the problem that needs to be solved, being able to make judgments based on both intuition and available data, and being able to evaluate and assess risks associated with the implementation of the proposed solution from different perspectives e.g. ethical and societal
No. The Artificial Intelligence vocabulary or terminology have been developed by the mimic of human intelligence and that is very limited in terms of understanding the very vast and complex structure of neurons.
Good question! Ultimatey this will depend on an unambigious consensus definition of 'intelligence'. If it is merely to provide a written answer to a written problem, humans and AI can both be measured against this criterium. The better the written answer, and the faster it is provided, the more intelligent the human or AI is. If our definition of 'intelligence' includes wider aspects like creativity, the ability to generalise learnt information to new/novel domains, emotional intelligence, physical interacton with the environment, accurate inference/prediction, lived experience in a body, short/long term memory, innate imperatives like own drives/needs, etc. we will have to define 'intelligence' more comprehensively. The reason that I am responding to your quesion is that I have an interesting, possibly unique, perspective to add. I have developed a type of AI that is not connectionist (i.e. not generative AI like LLMs). This different AI takes the form of a functional brain model, called the Xzistor Mathematical Model of Mind, that accounts for human-like intelligence (including cognition and emotion), and this framework needs to be instantiated inside a robot body (as we have done and demonstrated successfully). This type of AI, you can refer to it as symbolic and in future neuro-symbolic, uses the human brain and models it functionally. It has been proven to work well in both physical and virtual agents and I will now share with you some reading on this. Firstly, our latest paper is here and Appendix A explains the Xzistor brain model: Preprint Artificial Agent Language Development based on the Xzistor M...
I will also provide a very simple short book on how this brain model achieves intelligence by functionally mimicing the human brain Default Mode Network - allowing these agents to feel emotions, learn, think (solve problems) mentally and physically, and even dream. Here is the little intro book for free:
Preprint Understanding Intelligence PrePrint Version for Peer Review
I hope this will add some interesting perpsectives to your great question and lead to more discussion on this topic. If you need any more information, do not hesitate to reach out.
Rocco Van Schalkwyk, Bristol (UK).
- Badges
- Science topic