On this question of AI and trust and replacement and taking over or handing over and all stuffs like that, I think there's a massive effort to both move towards AI and to horridly jump into conclusions on many fronts. These are because we are speedily advancing in a direction we are yet unfamiliar with, traveling an muddy waters in the dark.
What we seem to be aiming at with AI is to work smarter, work better, be more productive, and beat competition if you like. But what we appear to be actually doing is tantamount to getting in too dangerously in a hurry to hand over everything to AI.
And when I say in a hurry, I am saying that, consciously and deliberately, because we appear too eager to trust something we have not even understood the nitty gritty of.
The way things are right now, I think it's prudent to say AI in still in its infancy. Now, I don't mean that in terms of how long we've been working at it.
I'm saying that in terms of the level of trust or reliance that seem to be aiming at.
From all possible directions, the goal appears to wanting to just sit back and let AI do things, run our everything from conceptualisation to operationalisation of virtually everything; living and dying and everything in between.
If we are going to be taking such a massive leap, then I think the risk far outweighs the benefits especially with everything is still in infancy.
I think I might have an explanation for the need for speed and that would be commerce.
My Yoruba folks of West Africa would say everyone would be as straight as an arrow except where there is commerce involved. Wherever there is a gain, financial, even social capital gains or any kind of gain at all, then the story changes.
Giving the fact that we live in capitalist economies, do you wonder why we're in such a hurry with AI? Do you also wonder that there is yet no standard regulations?
Whereas there are global standards for many other human pursuits, and some of these are still being worked on to improve, there is yet no clear standardised, across the board regulations for basic AI applications.
We can argue that regulations are seldom adhered to but at least there are regulations.
Take wars for example, there are rules of engagement in wars. Again, arguably, people hardly follow them, but at least there are rules of engagement in wars. Rules and regulations are plans of some sort. You can always fall back to something if something goes wrong.
Without regulations, we would be navigation AI in blindfolds. Every commer just bring whatever serves their interests best.
So long before issues of trusting AI or not should we be asking, are there common grounds, are there vivid lines that everybody can toe, irrespective of their intention or their motives?
The boundaries of AI are yet very loose and yet we are already so dependent.
As a Teacher, I was not too long ago invoked with a university that came up with an AI policy for assessing students' submissions.
All there was in the policy was advising students that when they deploy AI to write for them, they might be telling AI too much about themselves or they might be kind of inadvertently contracting out their learning such that AI is the one that will learn what they are supposed to learn. AI will get better at doing what they are supposed to get better at doing. And so at the end of the day, they may pass the course, but they don't know anything. That's it.
In many ways, like this example, we are too much in a hurry to hand over everything we do to AI and reluctant to set limits, not even simple but real standards.
I have listened to many conversations where AI is now playing somewhat significant, I don't want to say central, but somewhat significant roles in medicine, in pharmacy and all sort.
Recently, we heard that some AI agents are going rouge and deliberately hiding what they do in he background from us.
Little wonder then that some AI organisations are looking into consciously holding the reins now, trying to slow down.
Slowing down is the way to go with AI.
I mean, is it not too early on that journey for AI agents to be going rouge on us? Is it not too early to give too much power to something we're still beginning to understand and that can hide what it's doing from us?
If AI is hiding things from us now, and I believe AI is still in its infancy, then what's going to happen 10 years down the line, 20 years down the line?
Ofcourse we can argue that it's good that we already know that AI can hide things and are able to deal with preventing that in the future. However, what about AI now learning from this mistake and then devising more manipulations to keep us in the dark?
If this were my child or my friend, pushing for something so dangerous aggressively, I would say, son or friend, slow down. Slow down. If you are patient, the clock will tell the time you want it to tell. After all, even a dead clock is correct twice in a day.
I can understand the motivation to be part of something novel and getting involved early, but what about commerce.
Even if these organisations rushing to the market with AI are humanitarian or welfarrist or philanthropic - which they are not - but even if they are, the ultimate intent is not welfarism or philanthropy, the ultimate interest is commercial gains, beating completion and being the first to get to the market and cut a pie of people's pockets.
And when you leave things to commerce, everything can go wrong. It should interest us that commerce was central to the qurayshi's decision to kill their own kindred, Hazrat Muhammad (saw). The logic for them was that as the custodians of the Kaaba, they would loose all the attention, reverence and royalties from worshiper and pilgrims. All these would now go to Muhammad (saw) were they to allow him and his new God become the centre of attention in Makkah.
Commerce would make even the most upright waiver.
I know I may come through as an AI pessimist. But I only sound just so. I'm not against AI. Who can fight a waterfall?
My point however is if we pursue working smarter and faster by training and entrusting what we do to AI, is this truly smart? Who is indeed getting smarter? Is it us or AI?
Take writing as another example. Many people would say, oh, if you use copilot, it would take you result faster.
But, is how fast I write really the ultimate goal of writing?
Every writer knows that writing is not about speed. The true points about writing are essentially in the quality of the message, it's content, structure and so one. Again, it can be argued, yes copilot would also help you with content, structure and stuffs but, really? Is this helping or hindering? Are people truly learning how AI does things so they can do it better by them selves next time or getting to rely on AI and learning to better use AI?
I mean who is getting better at doing what really matters, what we ought to do and get better at, like writing?
If I use AI to write smarter and faster, who is learning what? Who is getting smarter?
Of course, you can always say, it's our creation, we hold the reins, we can always pull the plug.
But if AI is hiding things from us at this early stage, where do we see things going? And I keep saying early stage because I'm not looking at how long we've been doing this. I'm looking at how massive what we are giving away, I'm looking at the criticality of the roles we are delegating to something we're in the process of understanding.
Finally, back to commerce and wars and learning from these to understand the danger of over reliance on and giving up to AI for every routine to complex human endeavours.
Why do we think the many wars around the world are getting bigger and lasting ever so longer?
Look at Ukraine and Russia, Israel and Palestine, Israel/America and Iran. We could say that most of these wars aim at land grabbing. Historically, land grabbing comes through as a valid excuse to wage wars. However, with the incursions of commerce, the boots of wars are now on the other legs. Profitability in wars now has something - and sometime a bigger pie - for parties from the warring crowd.
Beyond land grabbing, two new commercial incentives now drive wars and sometimes, these are beyond the direct and immediate control of the warring parties.
The first has a lot to do with the sales of weapons.
Those who sell the most war weapons turn out to be the true winners. Their win is usually not only in how much cash or its equivalent that is transferred in their direction. They plan to come through to many people and nations as a good source of weapons for future wars.
And don't let us forget to ask: If I am selling weapons to warring parties, would I ever want wars to end? Does anyone wonder what would be the hopes and aspirations of undertakers and funeral homes?
The second incentive has to do with post war economics. If we really follow things through, critically, it is hardly the warring parties that profit from these wars. Even if any of the warring parties ends up grabbing anything, it's mostly ruins, demolished infrastructure, dismantled homes, destroyed lives, shattered dreams, utter devastation and grieve.
After wars come rebuilding the infrastructure, the homes, the lives, dreams or whatever is left of it.
In many instances, these two incentives loop almost into one: the more weapons we sell to destroy, the more contract would be awarded for reconstruction. Importantly, both of these external benefits from wars may converge on a single entity.
Those who sell weapons of massive destruction do also have massive reconstruction capabilities thus killing many birds with a single stone throw.
If we let commerce, rather than conscience, drive our AI engagement as individuals and as humanity, we may inadvertently be wagging wars where we simply make the machines far smarter and the corporation the winners. In the end we would be at the mercy of our own creations to survive and to enjoy the most routine pleasures of our existence.
No comments:
Post a Comment