Tasty: Why AI Cannot be Disrespectful

AI is NOT Disrespectful - It lacks the capability! Whether or not certain AI's may be clumsy (certainly) or biased (like all humans) they are incapable to being disrespectful - here's why!

It is a category error that we are not just prone to but being encouraged to indulge in.

It's a huge mistake to take what is a metaphor (and a deliberate framing of our interaction with the machine) for the reality.

This particular error has a name: Anthropomorphism. Yet it still is in danger of becoming embedded in popular culture.

Perhaps because the experts in AI themselves indulge in it. Whether this is because of hubris or just playing up to the myths bred in science fiction and popular culture, from Frankenstein's Monster forward, hardly matters.

What does matter is the fear and false expectations that this is palpably creating on a widespread basis.

Not to mention the massive distraction it's causing from the urgent problems AI is really creating right now!

Real Unicorns and General AI

The idea of 'general AI' is an extension of this problem. Like real, live, unicorns in the field only glimpses have been seen out there. In both cases reflecting the preconceptions and faulty perceptions that our psychology well understand flow from them.

In both bases perceptions are being influenced by hopes, fears or expectations - but when it comes to evidence, there is none.

AGI (Artificial General Intelligence) may be a goal but how realistic is it? Well standing back from the stories and myths embedded in our culture it's easier to see unicorns arising, engineered from some lab, that AGI.

For all furore' he enjoyed when exiting Google, ensuring his name will go into the history books as the 'Godfather of AI', Geoff Hinton is on record saying AI "should be thought of as an altogether?different?form of intelligence to our own."

He's right in this. The idea that a computer could calculate many of times faster than a human was an object of fascination - and one definition of 'smarter' in earlier days. Now it's simply accepted as a mechanical fact.

We've been carried away with the idea that idea that these machines can be 'smarter' than humans - potentially by several orders of magnitude. But we're not the first generation to be affected by this form of hysteria:

A mild mass hysteria fostered by these modern myths. But they do matter.

Not least because it invites the very error evident in imputing 'disrespect' to the computer - the AI, and in so doing imputing personhood and responsibility.

(Of course we know that the moment any related cases ends up in a court of law it is the human using the AI who'll be held responsible, because decisions there are based on evidence rather that culturally embedded myths).

Fun as the myths may be (and I've probably sworn at my PC as much or more than the next man over the years) - and as useful as the framing may be in making what really is a new kind of 'User Interface' it remains a mistake, and a dangerous one, to regard any AI as more than a form of 'Knowledge Processing' - reformatting and regurgitating the data it was fed.

Could AIs become super-human? Yes, of course - in many ways they already are. They're quick and tireless for one thing, and have their 'fingers' on a wider range of information that any single human can hope for, for another.

But that's not the same as agency or personhood. Glimpses of which, without any evidence, serve only to provide evidence of our imaginations and misperceptions.

Real Dangers

The real danger though lies in the distraction from the real issues of AI: The risk of a jobs apocalypse that badly needs attention, and new policy thinking right now!

Together the adaptations needed are not just in the educational systems (with great opportunities as well as risk) and security. Which is already being impacted as packaged AI tools are now widely available to hackers, leading to a rise in the level and sophistication of phishing attacks. But also in our culture and online capabilities, to deal with the proliferation and normalisation of 'deep fakes' etc.

I don't suppose the myths and mild hysteria are going to go away any time soon - or indeed ever. They're part of the human condition.

But we must stop allowing them to be taken so seriously as to distract from the real changes and adaptations to AI that we need to address right now.