No, AI is not smart

"No, AI is not smart".

We recommend this forum written by Santiago Schnell, academic manager of Dartmouth College.

His argument touches the core of what it means to understand.

We emphasise some ideas that invite us to reflect:

"Chaining reasons is not the same as understanding what they mean, why they come to the point and how far they allow them to go, or answer for the consequences of a statement."

"If we attribute intelligence to machines exclusively for their results, we can end up redefining human intelligence according to what they do best. A society that measures intelligence only by performance ends up impoverishing the judgement it needs to decide what to do with that performance."

According to Schnell:

Current models are formidable instruments of ratio: they organise information, complete arguments, identify relationships. But they have not shown to possess intellectus.

➼ One of the greatest risks at the cultural level: companies, administrations and media have incentives to reward speed, volume and formal correctness. And the time spent deliberating, contrasting a source or recognising uncertainty may seem unproductive.

Full article here.