We cannot buy what is not done
The danger in a society that has embraced unchecked capitalism is that it becomes ignorant of the limitations in attempting to use capital to address every problem. Capital itself is a tool for managing allocation not creation. You cannot buy the results of research that is not done. You cannot buy expertise that is not built. You cannot buy skills that are not developed. You cannot buy the wisdom and judgement that comes from experiences that don't happen. You cannot buy the understanding of learning that doesn't occur. You cannot buy institutional memory that is not shared when people are not there to share it. You cannot buy relationships that are not built when people are not there to build them. Capital doesn't create. It allocates. It appropriates. If creation is interrupted or ended, it cannot be replaced with capital. Capital can only allocate resources that facilitate creation.
Generative AI is an unsurprising thing to be enthusiastically and recklessly embraced by a capitalist society because it offers the illusion of replacing the creation of knowledge, expertise, and skills. However, as it can only remix and operate within the rules, patterns, pathways, and gaps that already exist, it cannot truly create. It looks like it can when what it's doing is more rapidly processing data in a way that's faster and more complex than is infeasible but not impossible for people to do when sufficiently resourced. New things aren't really being done so much as existing things are being done wildly faster with far fewer guardrails. It is still bound by the limitations of what has come before, even when those things appear novel when surfaced by AI products because we've yet to fully explore, identify, experiment with, or understand them. It is still coming out of the box that we packed even if we do not yet grasp the entirety of what we put in it. Anytime something goes faster than what has come before, people get a lot excited and a little bit stupid. We experience a phase of awe. However, excitement does not warrant abandoning reason and circumspection. Stupidity does not excuse recklessness. Awe is not a replacement for deliberate, intentional consideration of impact.
The advancement of AI technology does have utility, undoubtedly. But it's a mistake to misunderstand what the utility truly is. It can be useful in responsible hands guided by measured expertise under the appropriate limited conditions and contexts, but it cannot create or replace expertise. (And it's important to note there's a significant difference between domain-specific and purpose-built models and consumer facing products.)
The pressure of capitalism and the hubris encouraged by the excessive leeway and deference that's been granted to corporations, especially tech companies, have managed to confuse very important distinctions. People are mistaking being customers of generative AI products as being literate about AI.* Processing speed has been mistaken for intelligence. Quantity has been mistaken as productivity. Efficiency has been mistaken for effectiveness. However, none of these things will likely standup long-term under the true weight of the dependencies and obligations of our systems and the complexity of our societies. Worse, they already fail us in the increasing violations of our privacy, our freedoms, and our ability to trust our perception of the world around us.
The broader field of AI development increasingly represents a risk to much more as the unchecked race for industry dominance and the embrace of automations sends us surging past warning sign and after warning sign with no indication of slowing down to think, to implement guardrails, to abide by necessary precautions to prevent serious, even irreparable, harm.
Capitalism is a self-protecting system that will tolerate a limited population with a deep well of skills, knowledge, and expertise that it can commodify but it requires a broad population with limited knowledge in order to continue unchecked. The moment capitalists can replace the well of intelligence and skills or believe they can extract it and separate it from a human being that comes with demands and the ability to vote, then they will no longer indulge the continued development of skills and intelligence within the population and will treat it as a threat.** Thus we find ourselves hearing people say things like "We see a future where intelligence is a utility like electricity or water and people buy it from us on a meter," tech CEOs are lining up to discourage higher education after building their companies off the expertise of thousands of college-educated engineers and scientists, and the Department of Education is being dismantled. We find ourselves in a position where consumer-facing generative AI products likely represent one of the largest censorship threats in generations as corporations with deep political ties and economic incentives control what information is generated in response to prompts, biased automations reduce our agency, and accessing the open web suffers from the reduction of true search.
Right now we're in a difficult and consequential time in which capitalism-oriented businesses and capitalism-ensnared organizations and institutions are trying to leverage AI to replace skills, expertise, and labor that they either don't have the capital for or don't want to expend capital on. But we cannot buy what is not done and what is not done will cost us. The bill for the thoughtfulness, precautions, and creation of skills, expertise, knowledge, research, and experience that doesn't happen now will come due.
There is a world in which we can judiciously, intelligently, and carefully use these technologies to equitably improve the world for everyone but that requires prioritizing and protecting the creation of the things that neither capital nor machines can replace.
This essay includes some things I've posted online previously. Links for access to the originals:
*References a BlueSky post I've previously published.
**References a Threads post I've previously published.
