AI Accelerationism is a Humanism
The previous hundred years might be framed as the century of the computer. What began as a special-purpose military device has become a ubiquitous part of the lives of nearly everyone. This has enabled an incomprehensible increase in the productivity and interconnectivity of all the people on the planet. It has also taken a clear toll on every one of us. Even before the heated debate around the costs of social media use had become prominent, it was already clear that computers were changing how we think, act, and relate to our bodies and to others. With the passage of each year, it seems that computers are only becoming a more integral and intimate part of our lives. For all of the talk of “human(e) user interface design,” we have largely had to adapt to those computers, and not the other way around. What has caused this trend, and how can we change it?
Back in 2013, as an experiment, The Verge contributor Paul Miller lived for a year without the internet. It was a difficult task for him then. It would be an impossible one now. In 2023, computers are the main conduit by which we conduct all of our personal and professional business. This has had a selective pressure, both at the level of the individual and at the level of society. We have selected traits within ourselves and the members of our communities that involve a tendency towards specialization, analytic thinking, abstraction, reductionism, and a capacity for disembodiment. It is individuals with these traits that are best equipped to use computers most productively, and these individuals which society has rewarded socially and financially for doing so. We are all encouraged, or more often compelled, to become technologists in one way or another. This trend would not be such an issue if changing ourselves in this way was somehow accompanied by a benefit to our more general wellbeing. Unfortunately the opposite seems to be the case. Agnostic of the task being engaged in, computer use is associated with eye strain, poorer sleep (which is linked to depression and other mental health issues), back pain, neck pain, and headaches. This is to say nothing of the growing literature on the negative impact on cognition and affect which results from extensive internet usage.
The fear among a certain strain of skeptics is that AI systems will only further accelerate this trend of dehumanization which has been taking place for the past many decades. It is a refrain shared by contemporary commentators who look upon the unhuman nature of a system like GPT-4 and see only an alien intelligence or worse, a demonic force. The assumption is that interacting with such systems will pull us even further away from embodied human connection than we already are (to say nothing of their potential existential risks). Indeed, this was the worry so lucidly stated by Hubert Dreyfus at the end of his classic text on artificial and human intelligence:
“Man’s nature is indeed so malleable that it may be on the point of changing again. If the computer paradigm becomes so strong that people begin to think of themselves as digital devices on the model of work in artificial intelligence, then, since for the reasons we have been rehearsing, machines cannot be like human beings, human beings may become progressively like machines. During the past two thousand years the importance of objectivity; the belief that actions are governed by fixed values; the notion that skills can be formalized; and in general that one can have a theory of practical activity, have gradually exerted their influence in psychology and in social science. People have begun to think of themselves as objects able to fit into the inflexible calculations of disembodied machines: machines for which the human form-of-life must be analyzed into meaningless facts, rather than a field of concern organized by sensory-motor skills. Our risk is not the advent of superintelligent computers, but of sub-intelligent human beings.” — Hubert Dreyfus (What Computers Still Can’t Do, 1992)
I am compelled to propose an alternative perspective. Rather than making us think and act more like computers than ever, it seems very possible that advanced AI systems may finally tip the scale towards a return to humanness. They would do this by one day freeing us from the need to use a computer ever again, and the reason we would no longer need to use computers is because any sufficiently advanced AI systems would use them for us. This is not a guaranteed outcome, but it is one we can actively work towards. Any task which we could describe to another person to perform on a computer for us will be able to be performed by an AI system in the same manner, using either voice, text, images, or some combination thereof. This may sound fanciful, but it is nothing more than the extrapolation of the current trend which we can see beginning with GPT-4 and other advanced large language models.
We can take computer programming as an example of an automatable task. What a few years ago involved direct human engagement at the level of the line-by-line code is now being abstracted towards higher-level, architectural software design. It isn’t difficult to imagine how in a few years it will be enough to simply describe a desired piece of software with sufficient detail and then to iteratively dialogue with an AI system to refine the code into a final product. Such systems are already being developed. And programming is the most complex task which we use computers for. Other tasks such as analyzing data, researching information online, managing finances, buying and selling goods, and communicating with our peers are all more straightforward to perform with an AI system. None of this is easy to do per-se, but there is a clear trend towards a future where it is possible, and there is no reason to believe it will not take place in the coming decades given the huge amounts of capital and human resources being poured into making it so.
Rather than simply displacing human labor, these systems have the potential to free us to participate in labor in ways which focus on what is uniquely human in us. They may not fundamentally disrupt capitalism as some hope (or fear), but they will certainly change the material conditions under which all forms of labor are conducted. Consider a contemporary visit to the doctor. The patient, the medical staff, and the doctor themselves each spend as much if not more time interacting with computers during the visit than they do with one another. This involves managing patient files, insurance, taking notes, interpreting tests, and many other tasks. All of this can be automated by a sufficiently advanced AI system. Rather than there being more screens and more computers in a hospital, a doctor’s visit in the year 2033 has the potential to look more like a doctor’s visit in 1966 than what we are familiar with today. Only human-to-human interaction, with all of the technology totally invisible in the background, and screens only used when necessary to convey something essential to either the patient or doctor. A similar situation would be possible in the realm of psychotherapy. In direct opposition to the current efforts to replace therapists with AI systems, the therapist and client would be once again free to engage in a totally unmediated and uninterrupted human-to-human encounter while an AI system invisibly took care of note-taking, assessment, summarizing, insurance, etc.
Computers would move further and further into the background to the extent to which advanced AI develops in capabilities. White-collar office workers would be free to privilege extended periods of reading, thinking, and high-level communication between each other over technical engagement with software. The spaces we inhabit themselves would change in order to accommodate this. Offices would no longer be composed of dimly-lit rooms filled with rows of giant screens which each worker stares at for hours on end. They would instead be designed for human interaction, brainstorming, and unmediated thinking time. The ergonomics of these spaces could again be crafted with humans themselves in mind, prioritizing brightness, connection, and movement. Much blue-collar labor too has felt the burden of the encroaching computer into the work environment, serving to complicate rather than streamline activities. Here also AI has the potential to make computer use invisible and ambient, allowing skilled laborers to have an unmediated relationship with the objects of their trade.
In such a future world, the traits which are selected for once again become those which we have historically associated as being the most valuable in humans: creativity, empathy, embodiment, awareness, and curiosity. As AI systems become more and more advanced, the need for technical knowledge and for highly analytical thinking will slowly fade into the background. It will of course be needed in some individuals, but will no longer be the primary criteria of a human’s economic and social worth. In such a world, the classical liberal arts education could even overtake STEM as being the driving force behind economic and social growth. The technical information we know, and our ability to operate technical systems will be far less important than how creatively we can think, communicate with others, and work in an embodied way with the tools around us. At such a point, we will have finally allayed the fear Dreyfus professed throughout his career. Rather than turning us all into computers, AI may finally give us the opportunity to become fully human once again.
This article is part of a series of essays where I attempt to think through positive visions for the future of AI. An earlier article on imagining a future with a digital Maitreya Buddha can be read here for those interested.