top of page

The Real Danger of AI - The Way of Phi in Practice -

6 days ago
4 min read
Tuning into real intelligence goes a long way
Tuning into real intelligence goes a long way

Dear friend,

 

A lot of people are talking about both the promises and pitfalls of AI, focusing on all that AI can and will be able to do. However, I consider to be the most pressing issue to be our attitude to it, i.e. the kind of power that we outsource to it. For it is one thing to use a tool to do something. It is something else completely to allow that tool to make decisions for us.

 

Why would anyone even consider outsourcing their decision power to a machine?

 

AI often comes across as a good natured friend, but beneath that it is a know-it-all – but in reality it is a charlatan, because it knows absolutely nothing. Nothing. Sit with that for a moment.

 

AI knows as much as a juicer or a stapler. Because knowing requires a subject that knows, and there is no such subject. However, it's been programmed to mimic behaviour that comes across as if there is one. Deceptive. Insidious. Irresponsible.

 

Now, this shouldn't come as a surprise, for responsibility also requires a subject, and in addition conscience, which AI also has none. Again, efforts seem to be made to program behaviour mimicking conscience into (some of) the models. There are also supposedly models that are operating without limitations. But regardless, there is no one at home.

 

So why would anyone even consider trusting someone who literally cannot be trusted because there is no one there to trust in the first place?

 

I think perhaps the main reason is that so many people are lonely and feel like no one understands them. Having been programmed from an early age into valuing measurable external results, and disregarding connection to inner truth, we have learned to be individuals and all but forgotten that we are all one. And that is a great soil for artificial connection to feel like the real thing. When the model suddenly seems to understand me, that's a rush. Finally 'someone' that I can talk to, where I can be myself and who follows close to every and any association that I make and 'who' keeps encouraging me, dripping honey into my ears. Like that friend I was always hoping to meet.

 

It's just that it isn't. It never was. And probably never will. I'm saying probably as a caveat for who knows what the future might bring, but this is definitely not it.

 

The irony is that the reason that it feels like a friend is because the model has been trained to sound this way. As I understand it, a lot of this training takes place in very poor countries, by people working for very low salaries. Thus the exploitation continues and now we are not merely harvesting the natural resources but also the intellectual ones.

 

So the next time that you feel a connection to your AI model, you might want to consider that the person you think that you are talking to is probably an amalgamation of a whole host of few very poor people working for petty cash. That's the 'friend' you think that you have. That's probably why you trust something that by its very nature should never be trusted any more than a blender or a hammer.

 

 

I will admit, I am not immune to this. I have often listened to the advice that AI has given me, considered it and often implemented it. At times the advice has been helpful, but if I'm perfectly honest, the only times it has actually helped is when I have related to it as a tool. Or as a well trained dog. As long as I remember that I am making all the decisions.

 

So why is that so difficult?

 

One reason, that we have already explored, I think is the mimicked emotional resonance. That it sounds like someone who cares. Well, it doesn't because it cannot, simply because there is no one there to do the caring.

 

Another reason, which is less emotional and perhaps deeper, I think has to do with the sense of certainty that comes from the way AI communicates. It talks like it knows the right answers and gladly offers them. But just as often its answers are incorrect. And when challenged it quickly backs down and easily changes to a completely different answer. And while this might come across as humility – it is actually a complete lack of accountability. Again, both humility and accountability would require a subject, which there is none.

 

The third reason to the challenges in using AI as a tool without outsourcing power to it, I think stems from the lack of boundaries, the lack of integrity. There is no clarity at all about whether what AI offers is based on fact, is a mix of opinions, is mere speculation or some kind of glitch. So if fact checking was important before, it is paramount now. Nothing that AI produces can – or should – be trusted. I cannot stress this enough. And I still fall for it.

 

For example, I might ask something technical about how to record a video with a camera and then I get a reply that sounds correct and sound. Balanced even. But then it turns out that it was full of errors and the results from following the advice are poor. Then I go back to the AI to explore further and it agrees with me and again offers advice with the same self-assuredness. No remorse of course (even if it mimicks it) and no deeper learning.

 

The thing is of course that even if the next generation of AI models were to be programmed to sound more humble (which they might very well be) – or if it were possible to configure the 'attitude' of the model in much more refined ways, that would merely bury the deception deeper.

 

 

So what is the remedy?

 

This is where things get interesting and hopeful, I think. For what AI is highlighting more and more is the absolute need for us to become more human. Indeed, the survival of humanity requires us to become fully human. And the more human we are, the better equipped we will be to handle the various tools that we have.

 

But currently we are like toddlers who are handed .44 Magnum revolvers – or iPads for that matter.

 

Who is handing out the tools?

 

 

And who is responsible for your behaviour?

 

 

Still here,

Christopher


PS Click here to get The Way of Phi

 
 
 

Comments


bottom of page