How governments can ensure stability in an AI-driven labour market of the future.
In 1775, the invention of the steam engine sparked the Industrial Revolution, forever altering the course of history. Two centuries later, the rise of the internet brought about the Digital Revolution. In the past months, another major milestone in technological advancement has been reached with the birth of ChatGPT on the 30th of November, 2022. In just about six months, it is already being used to write homework, write and correct computer code, and draft legal documents. This glimpse into the future is fascinating. Yet, what precisely will the future hold? To gain a clearer understanding, it is crucial to understand that ChatGPT is only a steppingstone on a much longer path. Over the next decade, the capabilities of artificial intelligence (AI) will inevitably improve: what is now a mediocre high school essay will eventually be a cum laude PhD thesis; a few lines of Java code might be top-selling computer games; and why would it not be able to draft an entire constitution? AI technology has the potential to revolutionise the landscape of economics – to start a new era, an "AI Revolution".
For now, this is all merely a vision. As such, we do not know how reality will unfold. Instead, take this with a grain of salt. Trying to predict the future has historically been a rather shady business, therefore, toward the end of this article, I will examine to what extent this prediction is realistic. My focus, however, will simply be to explore the possibility of what could happen and what to do if it happens. Enough about technicalities, though; let's get into the fun stuff.
What Influence AI Could Have on the Labour Market
Artificial intelligence is being used in medical diagnostics, financial data analysis, or in the organisation of global logistics – as such, it is not just a recent development; it has already influenced the labour market to some extent. In contrast to these more covert changes, those that could follow its most recent developments promise to be much more ground-breaking than any before.
At the time of writing, OpenAI's GPT-4 is arguably the AI most "human-like" in terms of its cognitive capabilities. It can emulate numerous intellectual activities typically performed by humans: Be it essay writing, coding, poetry, or roleplay. Upon examining these abilities, one common thread becomes strikingly clear: they all revolve around the mastery of language.
GPT-4 is so good at this because it is a "large language model", an AI model trained through the input of large amounts of text data. While evidently other types of AI exist, language models seem to be most useful in aiding (or replacing) human jobs; their primary source of input and output is language, just like for humans. So what shifts can we anticipate from AI in this context?
As a general guideline, I propose that any job done mostly on a laptop, i.e., jobs of which the input consists mainly of text, will either undergo significant transformation (such as those in managerial positions) or face complete replacement (like that of copywriters). Jobs requiring heightened organisational skills, where the text serves mainly as an instrument for transmitting information rather than for the direct creation of final products, might – due to this added layer of complexity – enjoy some more resilience yet are likely to eventually be replaced too.
Other, more social jobs such as caretakers, psychologists, or babysitters are likely to persist much longer. These tasks rely on complex personal interactions, which are more challenging to automate. While the demand for more workers in these sectors undoubtedly exists, it remains highly questionable whether this sector could be scaled to accommodate everyone. A significant reduction in working time and working hours seems necessary. Regardless, the individual's contribution to economic growth is sure to diminish.
It can be useful to understand this development through the lens of "capital versus labour". In traditional economic theory, capital refers to resources used to produce more goods and services. Labour, on the other hand, refers to the human effort utilised in this production process. The effectiveness of capital is generally more expandable. Although labour processes can be improved, most major improvements in production efficiency result from improvements in the means of production – capital. Historically, capital has, therefore, more and more replaced labour. In the past, this labour replacement was generally offset by labour expansions to different, often newly emerged sectors. For example, as automation replaced jobs in manufacturing, it also opened new opportunities in fields like robotics and programming. Projected developments in capital shrink the space left for labour to expand into; with AI becoming more and more human, human utility will becomes less and less apparent.
What Are the Societal Consequences of This?
Assuming we accept the previous analysis and anticipate the possibility of most human jobs becoming expedient, it is crucial to explore the potential impact of this development on society. In a world where most human work is seen as redundant, will most human workers become redundant as well? In other words, how far is political empowerment possible without economic leverage?
In contemporary society, if the political system fails to serve the people's interests, the populace has the ability to withhold its support. People can resort to strikes or cease paying taxes. Modern dictatorships, despite their oppressive mechanisms, still rely on their working class. They might enforce stringent rules and regulations, but fundamentally, they must ensure the workforce remains productive. This necessitates providing basic necessities and maintaining a certain quality of life. In essence, modern governmental power is inextricably linked to the power of the workers.
However, significant technological advancements, such as the proliferation of AI, could potentially undermine this economic leverage that workers currently wield over governments. It is important to clarify that this scenario is not about AI conspiring against workers, nor is it coercing governments into disregarding their needs. Rather, the concern is that as governments become more reliant on AI and less on human labour, they may unintentionally become indifferent to the needs of the workers. This phenomenon is not a new one: there are countless examples of the economically weak being forgotten in the political process, be it during investments in infrastructure, education, or health care. The core difference between what we have now and what might happen in the future is the scale. Currently, the economically unempowered make up "only" a particular chunk of the population. These disparities will only be exaggerated in the society of the future. It could consist almost entirely of a class of "economically useless".
As we move further into this AI-centric future, navigating these shifting dynamics with care and foresight is imperative. Political inaction could inadvertently tip the scales, causing societal imbalance and unrest. Potential indifference of governments towards human workers, not out of malice but as a by-product of their futility, could cover them in a veil of political invisibility.
Thus, how do we approach this situation? How do we maintain a healthy balance between political empowerment and economic influence in a world where AI reigns?
Worst Cases and Best Cases of AI Implementation
The central question that arises when contemplating the implications of AI implementation is the classic socio-economic conundrum: who owns the means of production? The answer to this question could steer us towards varying scenarios, ranging from dystopian to utopian. We will explore three scenarios on this spectrum.
In the worst-case scenario, private corporations or individuals would own extremely powerful AIs that will, as previously discussed, disrupt the labour market. In such a situation, most of the profit generated by these AI systems would be concentrated in the hands of a small group of people. In the absence of effective AI regulation, this would leave the rest of the population without jobs or income. Such extreme economic inequality would be no doubt reflected in a nation's political processes. If we simply let AI happen without government intervention, dystopia seems imminent.
One possible mitigating approach would be to combine privately-owned AI with the concept of universal basic income (UBI). This idea first gained fame in Silicon Valley and through promotion by many tech billionaires. In this scenario, AI would remain privately owned but would be heavily taxed, with the proceeds used to fund the UBI. However, this approach still leaves the means of production in private hands and those living in society as mere recipients of UBI cheques. As portrayed in the previous section, a population without economic leverage is at risk of political disempowerment. After all, what can they do but accept the choices made by whoever owns the capital producing their cheques?
In an ideal scenario, we could envision a publicly owned, perhaps even open-source, AI. All profits and power generated by AI would be owned by the community rather than private entities. Each citizen gains an equal right to the rewards. Such a scenario has been termed "AI Communism" – the shared ownership of the means of production; of powerful AI. It promises stability and equality in an age of uncertainty and polarisation. This, however, would necessitate major shifts in our economic and political paradigms and socio-economic structures.
Looking into the future
Now, are these predictions realistic? Is “AI Communism” the solution? That is hard to tell. Personally, I remain unsure whether any such scenario could occur in the near future. Nonetheless, even if the likelihood of this new economic revolution is small, its potentially disastrous impact justifies the attention this topic presently gains. The following years will no doubt be interesting ones. The world could well change to an unimaginable extent – we should try to make this change be for the better. Whether we succeed at this, only time may tell.