In the 60s, self-service gas stations became popular in the US, marking a change from the old model in which attendants had to go out to refuel each customer. The small managers were dissatisfied: they could not afford the cost of adapting to self-service, and in any case they were not convinced of its effectiveness. After a few years, however, most gas stations had adopted self-service, except in some American states where managers protested with the support of local fire departments, fearing that untrained drivers would cause fires by overfilling. the tanks.
You know how it went: "human" service stations weren't as profitable as self-service ones, so the operators themselves pushed to abolish the bans. In 1992, 80% of gas stations in the United States were self-service. Today, self-service is the norm. Except in New Jersey, where a law still requires human service: thousands of gas station attendants still work to carry out tasks that are now automated everywhere. What does this story make you think? It makes me think of artificial intelligence, and in general of future time.
Towards the future time
The rapid expansion of artificial intelligence models such as GPT-3, Give her, StableDiffusion, MidJourney, Github Copilot and others is raising concerns among the public about the impact of automation. About who? Well, artists, designers, writers, programmers, lawyers, and workers in almost every other industry.
In the past, automation has often replaced jobs in specific sectors that required a lot of manual labor, mainly in the lower income bracket. Today, things are changing: AI systems are no longer limited to the automation of specific tasks. Furthermore, general AI systems are showing exponential growth in basic skills such as reasoning, writing, fact-checking, humor, and conceptual understanding of images and text. Quick solutions are emerging to some of the most common challenges in machine learning, such as difficulty in memorization and rigidity in problem solving. In the past it was necessary to provide a large quantity of examples to the algorithms so that they could learn, but new machine learning and engineering methods allow these systems to be "trained" very quickly.
But it's not just this. Some models are now surpassing humans in activities such as i SAT test, and standardized math tests. Chat GPT, despite being a model not specifically trained for the law, has recently passed a legal practice test. In the future, the cards will be shuffled for everyone, even for those who still feel "safe".
We are already on the bend
Our society has been gradually changing due to automation for some time, but what distinguishes this period is our position on the exponential curve. Progress is happening much faster than a human career and there is no sign of slowing down in the future. You can see it in this graph: AI models have doubled every 16 months since 2010, and this growth won't stop. It's not about years: it's happening now, and the decisions we make about AI now will affect its development for decades to come.
Technological evolution is changing our world, and we need to decide how we want this change to affect workers. In the not too distant future, perhaps within just a decade, do we all want to work in boring, unfulfilling jobs, earning only the minimum to survive? In other words: do we want to find ourselves like the gas station attendants in New Jersey, or do we want to fight for a society in which artificial intelligence makes us all more satisfied (and maybe even healthier)?
Don't think that our life without work as we conceive it today is "useless" or undignified. Personally, I am not afraid that artificial intelligence could "beat us" in many fields that concern work. Even in chess, computers have long been destroying human beings, but this doesn't stop us from continuing to love two chess players who compete with each other. Everest can be reached much easier by helicopter, but this does not detract from the beauty and grandeur of getting there by climbing it. In the future it will be the same for work, if we detach it from the idea that it must necessarily have a direct connection with bosses and salaries.
The problem is not AI.
Modern capitalism has brought enormous technological progress that has enabled enormous wealth creation (for a few). Even the advent of artificial intelligence is the result of capitalism, in the end. But if we talk about work and satisfaction for everyone, probably capitalism is not the right model. The real problem is not AI, but it is a system that forces us to "despise" those who need help, to consider it "fraud" to be helped by the state, or to live thanks to a universal income. In our current system, people are forced to work or face grueling consequences, even if their work makes no essential contribution to society. Of course, it is important to remember that sudden changes are not feasible and would not be desirable in any case: but they must be introduced, a direction must be chosen.
Which direction? One possible would be one that passes through a high and progressive tax on large companies that profit and will increasingly profit from the use of artificial intelligence. The funds raised with this tax could be used to finance a guaranteed income for those who earn less than a certain threshold. In this way, even those who are unable or do not wish to work "as people work today" would have the possibility of having a secure income and dedicating their time as they wish. This is the future we must consider.
