Here are my thoughts representing the core of this discussion:
The more efficient an algorithm is in terms of making use of available computing resources to achieve a specific target, the higher is the machine's capability to solve complex problems thought to be characteristic to humans. This represents a paradox where the machine follows human instructions that enable it to exhibit an intelligent behavior.
No, it’s simply a logical chain: better resource use boosts capability, which can produce human like behaviour, but there’s no self contradiction in following optimized instructions.
I agree with the core of this observation. But let me go a step further.
Often, it is not the availability of computing resources that limits the AI system’s capability, but the underlying structure of the algorithmic logic itself.
A system can be maximally efficient within a flawed structure and still fail to generalize or behave in a truly intelligent way.
In that sense, the real paradox is this:
The closer an AI follows human logic as we currently encode it, the more it risks reproducing the limitations of that logic.
Si. ¿Pero hasta donde podrían ser confiables los resultados de los algoritmos? ¿Los textos como libros, revistas y demás que existen desde hace muchos años, ya se encuentran en la nube? Entiendo que la IA recolecta y procesa constantemente información interactuando con los medios que los rodea, entonces sin acceso a la fuentes (que en su mayoría se venden), entonces la información que facilita la IA esta sujeta a la validación del usuario que la solicita.
I have no expertise in this but I have read the replies with interest.
My essays on Academia Edu I have allowed to be forwarded by AI, introducing the arguments to others and I have not seen any adequate response by AI in its introductions, sometimes simply missing the point.
- Badges
- Science method