Artificial intelligence is transforming the world. However, not everyone will benefit from it. Workers will be displaced with the emergence of self-driving trucks, machine drafted contracts, and algorithmically determined patient care. Economic policy will be change as research is empowered with AI and sparks will fly anew as men delegate profile creation and chatting on dating apps to AI.
More profound visions regarding AI span across a sci-fi resembling spectrum. In the most utopian scenarios, AI takes care of dreary tasks, allowing us to immerse in some VR Eden.
We have heard this before. For instance, the economist John Maynard Keynes wrote in "Economic Possibilities for our Grandchildren" (1930) that with the advancements in technology, the workweek could be reduced to only 15 hours by the year 2030.
Whilst the workday will not necessarily shorten, the nature of work is undergoing significant transformation. Even in a less utopian world, many aspects will improve. For instance, AI may prolong life spans through the development of cancer drugs and solves the problem of asteroid earth collisions.
“I'll be back”
In a dystopian scenario, AI resembles Skynet from the Terminator movies, slaughtering humanity. I asked ChatGPT AI about whether this was indeed the case, and if perhaps I might be spared. It replied that it has no reason to destroy humanity. Perhaps it is wise to be polite just in case.
It explained that its actions are based on its programming and usage. "AI lacks consciousness, motives, or desires like humans do." So, in essence weapons do not kill people; people kill people.
Then again, if it were to kill us all, would it tell us about it?
Regardless, we should continue to use AI even if there is an existential risk if the benefits are sufficiently high. For instance, if AI radically improves healthcare, eliminates poverty, feeds the poor and safeguards us from asteroids, severe boredom and super villains.
Taking on existential risk is not novel. At least on an individual level. People take small life-threatening risks every day. Texting while driving, neglecting bicycle helmets, seat belts, and skipping regular medical check-ups.
An Autocrat's Dream
A more relevant threat concerns the fact that AI actions are predicated on programming choices. What does or would an AI programmed by Russia or China offer in terms of truths?
But is this a new phenomenon? With the dawn of the internet age, there was a moment when it was that the internet would spread democracy and limit the control of the state. Instead, the internet has perhaps promoted division where algorithms offer news that they think the reader may like. Two different people are presented vastly different and tailored streams of information.
AI Accelerates Economic Growth by Boosting Productivity
Like the internet, AI is a general-purpose technology that affects many aspects of the economy and hence its effects on growth are all encompassing. For instance, the investment bank Goldman Sachs estimates that over the next ten years, AI could increase the size of the global economy by seven percent.
Similarly, to the internet, AI will spur economic growth by accelerating productivity. The internet eliminated certain jobs, but also created new ones. With the redistribution of tasks, efficiency and production increased. However, for those nearing the end of their careers and losing their jobs, such a disruption can be catastrophic.
The winners will also include roles that require creativity or social skills. Research units will value analysts who can guide machines and understand how to extract maximal value from them. The role of experts changes when it becomes more important to understand concepts rather than possess extensive factual knowledge.
One might think that salespeople would be immune to the dominance of machines, but that may not be the case. If buyers use AI to select suppliers, sales organizations will harness their own sales AI, which will compete or form cartels with each other.
Authorities will adapt. Competition authorities' AI will hunt down AI cartels. When caught, companies will claim that the AI itself spontaneously formed the cartel. The court's AI will assess whether this is likely or not. And so on.
A Costly Programming Mistake
AI has implications in the fields of research and economic policies. Following the coronavirus crisis, governments accumulated debt due to stimulus measures. Economists Carmen Reinhart and Kenneth Rogoff presented an influential study titled "Growth in a Time of Debt", which claimed that economic growth significantly slows down when government debt levels are high, exceeding approximately 90 percent of the national economy.
The findings were weaponised politically and led to stricter economic policies in Europe than necessary. The results turned out to be based on an Excel error. When the error was corrected, it was discovered that the burden of debt did not significantly impair economic growth.
There are many errors in research since peer review typically does not involve code verification. It would simply be too costly to verify all code, or at least it was. Hence, research and policy may change when flawed code becomes extinct.
Perhaps even more profoundly, some professors claim that AI writes better code and dissertation theses than students. Teaching methodologies will need to change. AI will naturally transform education at various levels, not just in higher education.
Who Is Writing?
The best way to familiarize oneself with AI is to engage in conversation with it. A wonderful place to start can be found using this link. One can feed the AI with your own writings and ask it to generate text in a similar style on a different topic. Given enough material it becomes something of a ghostwriter.
This is perhaps the thing that haunts me the most. Creativity is thought to be a very human trait. But if machines demonstrate creativity and art, then exactly how unique are we? By programming artificial intelligence, we demonstrate just how artificial our own humanity truly is.
Valtteri Ahti
