Artificial intelligence will destroy the world as we know it… – 02/17/2023 – Sidereal Messenger

Artificial intelligence will destroy the world as we know it. If we have intelligence and sensitivity, it can be refounded as a better and fairer thing. Which is by no means guaranteed – at the moment, we go with our eyes closed towards the abyss.

It’s been fun to see the reaction to the release of ChatGPT, the natural language model from US company OpenAI. There are the non-believers, betting that it’s just a fancy “lero-lero generator”; the worried, considering it a job killer; and enthusiasts, celebrating it as a productivity tool. All of them are right.

The model learns to speak by consuming tons of text and developing original responses from it, many remarkably skilled, some missing the mark entirely. So far, none of this is different from being human. We learn to write (not to say think) by imitation, and we are not versed in all subjects either. The main difference is that the Dunning-Kruger effect is more pronounced in ChatGPT than in the most egomaniacal human being: the AI ​​thinks it knows a lot more than it does.

Despite this, claim its appreciators, even with its current limited capabilities, it is already capable of streamlining many tasks previously performed by humans, freeing them up for more creative work and increasing their capacity to produce. Absolutely true. Ask ChatGPT and he’ll tell you that himself.

It turns out that the demand for work, even in the most bloodsucking companies, is finite. If it increases the productivity of some, the company can lay off many others, since, ideally, the amount of work will be determined by maximum demand and not by the total capacity of the employees originally employed.

There are those who say that a tool like ChatGPT will never reach the prodigy of truly brilliant human beings. Let’s assume for a moment that this chauvinism leads to a truth. And? AI does not need to be an Einstein or a Mozart to successfully perform many of the intellectual professional activities performed by humans.

And even if it were, we are on the verge of seeing it happen. In 2012, I interviewed the American technologist Ray Kurzweil for this Sheet, and he predicted that AI would reach the human level of intelligence in 2029. It is 2023 and anyone who has seen ChatGPT operate cannot find the projection an exaggeration. Even more in light of the consecrated Law of Moore, formulation of Gordon Moore, founder of Intel, who noticed that the evolution of processors leads them to double their capacity every 18 months, approximately.

This law is still in effect. Which means that in 18 months we will have potentially twice as good ChatGPT. In 36, four. In ’54, eight. The “spark” is already there for all to see. Now it’s a case of sitting back and waiting until (quickly) machines surpass humans in intelligence.

When they beat a chess champion, we were impressed. When they mastered games with more nuance and less math, like poker, we started to worry. So they mastered natural language games. Today there are AIs capable of passing tests for the practice of medicine and law. They already have a level similar to that of good doctors for analyzing image exams. NASA is using it to design spacecraft. Even computer programming is routinely and competently done by computers themselves. ChatGPT is just the layman’s gateway to a disruptive tool introduced in the world of work.

There’s already AI everywhere, and it’s only going to intensify. Humans will be replaced by computers, including in many jobs hitherto considered untouchable. (I was envious of colleagues who were recently able to write articles for the Sheet commenting on predictions and assessments that would have been right and wrong, so I’ll leave this one here, to have something to write about in another ten years if the world surprises me and takes a different path.)

Does this mean that all jobs will disappear? Will we live in a world where everything is produced by AI? No, there is no chance of that happening. There will still be plenty of room for human activity, and even in situations where the machine is predominant, a human being will always be needed to “supervise” and “sign under” so to speak the work produced by the black box. from AI.

In many circumstances, it simply won’t pay to switch from human to machine, given the human’s versatility to perform many tasks that involve a combination of dexterity and creativity. In others, it won’t make sense to change. No one will want to see an AI performing at a rock concert or playing in a football match (we’ll even have things like that, but more as curiosities and additions than as replacements). Everything that involves some level of empathy will still privilege the human.

However, it turns out that it is not necessary to eliminate all professions, or indeed any profession, to collapse society as we know it today. All you need is to cut a high percentage, say 40%, in all careers. Ready. We will have a colossal mass of unemployed people and, worse, psychologically broken.

Our culture is heavily built around a myth (now largely false) that tireless effort, intelligence, and productivity lead to success. Perhaps it was true when we were all hunters and gatherers. Those who hunt more and better have more to eat. As societies grew in complexity, many other factors came into play. Today, our social structure imposes far more conditions for success or failure than mere individual effort. It is much easier to succeed by inheriting privileges than by starting from scratch, and the social mobility numbers demonstrate this. Social policies are created to try to alleviate these competitive disparities.

However, the myth of meritocracy persists. Everyone remembers the story of the guy who left the bottom and became rich. The press always celebrates cases like this. Some version of “Silvio Santos was a street vendor and became a television owner”. It is rare for anyone to mention the fact that all the other street vendors in Brazil never became Silvio Santos, and certainly there are many who have worked hard all their lives.

With the arrival of a modestly capable AI (let alone a super-intelligent one), sustaining this crude notion of meritocracy will go from malicious to impossible.

THREE PATHS

Faced with the imminent arrival of the “technological singularity” (as Kurzweil and other scholars refer to that moment when a creation of ours appears that is equal to or superior to us), there are only three possible paths: the collapse of civilization, a banishment of AI or a post-capitalist minded future.

At the moment, we are firmly on track 1, civilizational collapse. The signs are out there. Economic stress and a brutal increase in inequalities, after a long time of reduction (on a global scale), a crushing of the middle class everywhere and pointing fingers everywhere, expanding xenophobia and extremism. The introduction of modicum amounts of AI into our lives (in the form of social media) is already powerfully disrupting us. And the gradual structural unemployment oppresses. The philosophy of “eternal economic growth” on which capitalism and its myths (including meritocracy) are based begins to run into limits of different dimensions: population, climate, environment, availability of natural resources. Some are circumventable. Others don’t.

And, of course, what continues to dictate the progress of society is efficiency gains. And that’s okay. This is how it has to be if we aim for development. If an AI can help clear the SUS surgery queue, allowing doctors to see more people, isn’t that desirable? It is more than desirable, it is an ethical imperative. Companies will be guided by the same logic: it is possible to be more efficient – ​​and therefore increase profits – with AI. Nothing wrong with that. It’s progress. But if not done thoughtfully (and it’s a fact that there isn’t much debate about this), civilization can collapse under its own weight. With a finite demand for work (say, the end of the queue at the SUS) and an explosive gain in efficiency, we will have tons of unemployed people who, unable to support themselves, will live resentful and susceptible to radical movements.

We need to get away from that route. As? Two possibilities, one of them not very credible: the banning of AI technology.

It only takes a moment to remember how impractical this is: how successful are governments in blocking access to the internet or specific content? There is a lot of effort, but success is doubtful. And trying to turn AI into a clandestine tool would be even worse, socially speaking, than encouraging it. In addition to creating more inequalities (between those who gain access illegally and those who cannot), it takes any opportunity to regulate and control its uses out of our hands. (Have you ever seen anyone taxing drug trafficking?)

The only really viable alternative is a reform of one of our founding myths, the notion that the function of work is sustenance. We need to change our mindset and understand that sustenance is a right, not something that is conditional on individual productivity. A full and healthy existence throughout life should be guaranteed to all human beings.

Hence, even among technology gurus such as Bill Gates and Elon Musk (to name two who are antagonistic in so many ways), there is a consensus that the future will have to pass through a robust program of universal citizen income. (Yes, Eduardo Suplicy was way ahead of his time, and we are lucky to have him with us.)

Two legitimate questions emerge from this: where will the money come from? And, solving this, will we be an alienated population with 90% unemployed?

The first answer is, I believe, a taxation system that manages to convert what high technology produces in terms of inequalities into profitability to sustain the program. In short, it will be marketers and consumers of AI services who will foot the bill, by making everything else in the system more productive but less human-intensive.

And what about occupation of the population? Work will take on new meaning and will be, rather than a means to an end, its own reward. The pleasure of creating, producing, realizing and studying can be exercised as a freely accessible privilege, instead of a duty linked to survival. Humans will be able to dedicate themselves to cultivating body, mind and spirit in the way they find most satisfactory. Will there be those who will do nothing? There will be. But if we assume that subsistence is a right, this is not even a problem – except for themselves, who may have a less satisfying life intellectually and physically. At the other end, there will also be those who, like today, will have great ambition, will want to work to get rich and will have to innovate to get there – the only news for them is that they will have to pay their contribution to others, in proportion to their success.

Brazil seems remarkably ahead in this transition, at least with regard to the aspirational aspect. Our Federal Constitution already guarantees a series of rights and we have a national income distribution program in operation, which can be expanded over time.

It will not be easy to keep up with the speed of this transition to the technological singularity, which promises to be fast and overwhelming. More easily implemented transitional alternatives will be to gradually reduce the workload of professionals, without loss of income, as AI gives them more productivity. That, from the outset, would allow the smoothing of structural unemployment, while we do not properly equate a universal citizen income program.

What will not be possible is simply not to reflect on the matter and pretend that everything is fine. To the enthusiasts of unrestricted economic liberalism, a message: no, the market doesn’t have the slightest chance of self-regulating in this regard. Doing nothing about these changes that have already reached us for some years and tend to intensify in a radical way will be an option every day less acceptable – and more dangerous. The future is right there.

Follow the Sidereal Messenger on Facebook, twitterInstagram and YouTube