Yesterday, I woke up to a passive-aggressive message written to me by my refrigerator. My cleaning supplies decided to unionize against me, and my half-charged telephone demanded maternity leave. Perhaps, this is a ridiculous statement, but does it not seem that technology is creeping nearer toward overtaking humanity each day? It begins with the fridge telling you to use oat milk rather than 2%, and before you know it, it is chairing board meetings, heading law firms, and doing open-heart surgery whilst simultaneously making fun of your midnight snack session.

These sci-fi scenarios appear to be fantasy, and to most of us, they may seem impossible. However, that is precisely why Artificial General Intelligence (AGI) is something to fear.  AGI is no longer limited to spitting out recipes or memes like the chatbots we currently interact with; rather, it concerns the possibility of machines that can think, learn, and adapt like people and maybe even be better than people.

We already can see the contours of this change on the work market. According to the projections of the World Economic Forum, by 2030, AI will generate 170 million jobs and will replace 92 million. On paper, that's a positive. Realistically, it amounts to a rearranging, one which  heavily relies on individuals who have prior access to high-level training in technology. The new jobs are clustered in the most elite engineering fields and the millions of middle class jobs in law, finance, and medicine. The occupations that had previously been believed to be immune to automation are now on the chopping block. For example, AI is creating demand for roles in machine learning engineering, data science, and cybersecurity, while automating tasks once done by paralegals, financial analysts, and even medical diagnosticians.

According to the Brookings Institution, AGI endangers intellectual work, which is the basis of white-collar jobs. Consider that: years of education, career experience, and culture can be overshadowed by an algorithm that never sleeps, does not require benefits, and is not afraid. In practice, this means that even professions built on advanced degrees and specialized expertise such as accountants, lawyers, researchers, and consultants risk being displaced by algorithms capable of performing their core functions faster and more efficiently.

The social cost of such displacement is enormous. Already increasing economic inequality would further increase with the concentration of wealth in the hands of those who own and develop such systems. The one percent profit as the rest re-equip themselves or face obsolescence. Democracy is weakened by the extremes of inequality, as in studies it is always seen that when the people feel locked out, they lose interest in participation. 

It is not a domestic issue. In the U.S. in 2019, there were more than 2.5 million immigrant STEM workers, many of whom have left their lives abroad to pursue opportunities in areas that AGI is now threatening directly. When even these skilled jobs are absorbed by machines, the shockwaves spread over the planet, leaving not just vacant pockets in the economy but shattered expectations of mobility and stability.

And as we move toward employment, a second dark side is the environment. Already, over 3% of the global electricity is consumed by data centers to power AI, and forecasts indicate that it may increase to 10% by 2030. The water usage is astronomical as well—some reports indicate that ChatGPT would need half a liter of water to answer only a handful of prompts. Scale that to billions of such interactions, and AGI is no longer competing with us for work, but it is competing with us for resources.

Yes, it is funny to picture my Roomba bargaining over dental insurance. However, behind the absurdity, there is a serious fact: AGI is not another device or program. It is one form of force that will be re-writing economies, societies, and even the environment. The question is not “Will AGI take jobs?” The question is “How will we shield people when AGI takes their jobs?”

Information retrieved from Dataversity, Piedmont Environmental Council, Reddit, and Linqto