How AI will destroy us
In fiction, we typically see AI destroying humanity through physical acts as in the movie, The Terminator. Unfortunately, science fiction such as this has blinded us to the truth: AI is getting ready to destroy us, but in a far more subtle way. Namely, it will do so by removing two fundamental ingredients from human society: our feeling of purpose and our need for other human beings.
Let us examine how AI will destroy us if we don’t act now, and what we can do to avoid it before it’s too late.
AI will remove our sense of purpose
The first ingredient is that AI will remove our sense of purpose because at some point in the near future, it will do almost everything better than us. Even if there remains some domains in which humans can outcompete AI purely on skill, AI simply needs to do a large percentage of tasks almost as good as humans and more cheaply than humans in order to take most jobs away from humans.
Some people give the rebuttal that if AI takes away our jobs, we can just do things for fun such as hobbies and retain our purpose that way. However, this does not negate the fact that most people will feel purposeless, for two reasons. The first reason is that many people find purpose through their job and will have difficulty finding purpose elsewhere. The people that argue that we can just shift to hobbies are typically highly intelligent intellectuals who are amenable to spending their entire life steeped in hobbies. That’s not a bad thing, but not everyone is self-directed like that.
There is another reason why we cannot switch full-time to hobbies for a sense of purpose. And that is, many hobbies will lose their appeal because most pursuits have challenges in them that AI will overcome more easily. I have already heard one person lament that they felt depressed about photography because they thought AI-generated ‘photos’ were better than their own.
Another classic example is chess. I love chess, but chess programs have made the game significantly less interesting, and a chess program is not even a very advance form of AI. Now chess is rife with cheating, and people use computers to make the study of chess so mechanical that it has lost its former magic.
The world champion chess player Bobby Fischer already felt this even before AI, and so he invented Fischer random chess, saying that:
I love chess, and I didn’t invent Fischerandom chess to destroy chess. I invented Fischerandom chess to keep chess going. Because I consider the old chess is dying, it really is dead.
He said this referring to the intense devotion of master chess players in studying opening book moves. Nowadays, all aspects of the game are so thoroughly studied with the help of computers, that AI programs such as Alpha Zero have truly removed the magic of the game.
This general phenomenon will happen with almost all hobbies. AI editing techniques in photography and AI recommendation algorithms are making photography more and more mechanical. Even the act of sharing your hobby is becoming more difficult, because AI algorithms are training people to merely follow trends, destroying the organic joy coming from sharing and discovering together.
In creative arts, people will feel a lack of purpose when they see an AI creating better art, or better stories than they can. Even if AI does not strictly create something better, it will create something sufficiently good with such ease and in such quantity that most people will have trouble finding human-made items or artistic creations.
Even if you can write a better book than ChatGPT 10.0, hardly anyone in the future will pay attention to it because it will be like getting your hand-crafted product sold on Amazon and competing against the elephantine mass-production machine of China. AI is already writing travel books en mass, and although they aren’t as good as the human versions, they soon will be.
Finally, even if you can somehow find purpose in a bunch of hobbies, how will you pay for them? AI will take over your job and thus the world will have to switch to some sort of universal basic income. But if you think that everyone is going to get a nice, equal slice of the pie, you’re mistaken. Those in control of AI systems will get the majority of the wealth, and you’ll be left on a cushy version of welfare, controlled by the elite who in turn control technology.
Well, perhaps people can still find some meaning in new jobs created by AI? Sorry. It is a fallacy to think that because new technology has created jobs in the past, now again, new jobs will be created by AI. A few new jobs will be created, but unlike in the past, far fewer will be created than necessary to support most people seeking a purpose. And even now, you can see that most intellectual jobs are becoming more and more specialized to the point that we are slowly turning into dim-witted monkey maintainers of extremely advanced technology.
AI will reduce our need for other human beings
The second ingredient is that AI will remove our need for other human beings. It will do so by making available almost everything we need without the need for human interaction. AI may not do so perfectly, but it will do so well enough that we will not have enough human interaction to develop genuine human bonds.
Some people argue that AI will never replace a true human being when it comes to friendship and love, and I agree. However, this does not negate the extreme danger that AI poses for human relationships. Why is this? Again, there are two reasons.
The first is that friendship and love is partially forged by truly needing another person to survive. The best friendships and relationships are formed by people coming together to help each other. With AI getting close to satisfying almost every need of almost everyone, the deep feeling of really needing someone will mostly cease to exist.
We care for each other because care builds up through a slow process of trust and helping each other, and that process will be broken by AI. Thus, we will become a world of narcissists, caring only about ourselves. Just ask yourself: how have you developed the strongest bonds in your life? It’s because you needed another person.
That AI will remove our need for others is not a wild prediction as much as it is an extrapolation of an existing trend. Technology has already partially removed our need for personal interactions with others, and even when we do interact with others, it has become far more mechanical through social media and chat apps. However, so far it has only removed a relatively small percentage of that necessary interaction compared to what AI will do if it’s not stopped. The effect that AI will have on how we need each other will be scary and largely unpredictable, but certainly horrible.
Other effects of AI
I already outlined the removal of purpose and the removal of our need of others as two of the main fronts on which AI will destroy us as a society. If that weren’t enough however, there are all sorts of other negative effects that will only enhance the devastation that AI will wreak upon us.
For example, AI has already been used in chemistry and medicine to discover new drugs and molecules. In a Nature Review piece, Jayatunga et al. wrote that
AI companies, most of which started less than ten years ago, have achieved a substantial fraction of the preclinical output of the top 20 pharma companies. We also already see examples of novel chemistry for major targets and a potential early glimpse at increased molecular diversity targeting new biological mechanisms.
This sounds good…at first. However, it is far from good. Just as researchers can use AI for novel drug discovery, so can malicious actors use AI to develop toxins, biological weapons, new viruses (COVID combined with Ebola perhaps?), and other dangerous things.
Then there’s advertising. Campbell et al. in the Journal of Advertising concluded that
Manipulated advertising is becoming an increasingly prevalent phenomenon in advertising. Techniques such as deepfakes and GANs leverage AI and machine learning to generate convincing, true-to-life synthetic ads that can be nearly impossible for consumers to detect.
If you thought advertising was bad before, think again. Now that advertisements can be made orders or magnitude more convincing and manipulative than before, we are headed for a world of intense short-term consumerism that will certainly do even more harm to the environment.
Some argue that dangerous medical research and malicious advertising were always part of the game. But those people are missing the point that it’s the speed that matters: by using AI, these dangers will hit us with the speed of a bullet train, whereas before they were merely slapped in our faces. We simply do not have the ability to cope with the speed of AI development of harmful technology.
If AI is so bad, why are we using it?
If AI is truly such a monster, why are we even developing it? There are two reasons. The first is that the initial stages of AI are relatively easy to control, and the most dangerous effects of AI are still some decades away. Just as with climate change and fossil fuel use, we are mostly ignoring the problem because the worst effects will be felt by our children rather than by us.
But there is another reason: short-term profits. Programmers and tech companies stand to make a massive short-term profit now. There is no regulation and no oversight to consider the long-term harm of AI, and tech companies are taking advantage of this. They’re taking advantage of us. Companies like OpenAI are completely immoral and are using the very system of our consumerist, capitalistic society to make a quick buck and leave us in the muck.
Finally, individuals use AI because it gives them a short-term advantage over others. Just like every other technology, it saves time. What’s the harm in a little automated photo editing, a little automated text preparation? But that’s short-term thinking. AI helps us do our jobs now, but we are training it to do our jobs. AI is making some of our tasks faster now, but soon it will take over those tasks completely. AI is amusing, but we don’t see that its giggle is the sadistic cackle of a mass murderer poised for a slaughter.
What can we do?
The dangers of AI are certain, and our fate down the AI path is unequivocal. Part of how we will suffer is known, and the other part is unknown but even more frightening for its unpredictability.
Given this, we really have only one option: destroy AI. Halt AI research. Of course, companies are now making so much money with AI that it will be hard to stop them. Thus, it’s in the hands of us, the people. Do your part by rejecting AI. If you own a business, keep it human and proclaim loudly that you will not use AI. If you work in an enterprise, do your part to keep AI out of your organization.
Some people believe that we should develop AI and use it for good, but it is obvious that we do not have the wisdom to use it in such a way. Telekinesis might also be used for good if we had it, but I would not want to live in a world where people could kill by thought. We are too primitive, and we have too many short-term instincts from the times of scarcity when we were hunter-gatherers. These short-term instincts were once adaptive, but in modern times, these instincts combined with technology and AI will produce devastating results.
Thus, caution is required. Instead of diving wholeheartedly onto the AI bandwagon, please, let us be cautious. Before developing AI, let us slowly restructure society so that we can be more mature in our use of technology. Let us slowly develop the wisdom to decide whether any new technologies are good for us. Let us not act as children, eating all the chocolate, for the following stomacheache is not one for which we are remotely prepared.
Conclusion
What can we say from these two points? AI will make us into purposeless, narcissistic and unhappy beings who will lose what it means to be human. We will become so confused and insane that we will not have the capability to go back.
And that is almost surely how AI will destroy us. Not by physical acts, but by removing the very essence of our humanity. We still have time to act, but that time is running out.
Originally published at https://blog.jpolak.org on July 30, 2023.
