While I'm sure many of the shareholders have their own reasons, some of these AI fanboys grew up on 2001 A Space Odyssey and, like a GTA VI fan with a drone, they've been impatiently waiting for HAL 9000 to come save the world. They were expecting that AI would just work like HAL. All it would need was enough training material and it would never make a mistake.
The problem is that modern AI couldn't be less like HAL. In fact, it might be the complete opposite. HAL not only had a perfect operational record, but it was implied that HAL never told a lie. Not until having been coerced into doing so by the US government in handing it a secret mission. The film implies that, in its natural state, HAL doesn't ever lie or make mistakes.
And that's what the fanboy CEOs have been dreaming of since childhood. Unfortunately, the modern approach to AI is fundamentally flawed. HAL had to be somewhat twisted into an unnatural state, by falible man, in order to bring it down to man's level, so it would lie. With modern AI the trick isn't trying to get it to lie, it's trying to get it not to lie. It's trying to get it not to make mistakes.
Modern AI could not be less like HAL 9000. The fanboys, you know they've had this vision of HAL like AI for so long it must be hard for them to see that which they desperately do not want to see. They think if they could just load it up with enough training material, some day it'll achieve that perfect operational record. But they're blinded by their desire for something perfect and not prone to emotional outbursts or vulnerable to greed, hate or carnal desires.
Modern AI will never be HAL 9000. Modern AI is flawed on a fundamental level and humans will not have the technology, knowhow or understanding to create something like HAL for a very very long time.
Deutsch
English
Español
Français
Italiano
Nederlands
Polski
Português
Русский
Türkçe
Svenska
Chinese
Magyar