We’ve got to stop calling AI companies “labs.” They’re companies wearing “lab” as sheep’s clothing.
Others have said this before. This post gets to the point quite well:
Continuing to call them labs makes them sound like harmless groups focused on pushing the frontier of human knowledge, when in reality they are profit-seeking corporations focused on building products and capturing value in the marketplace.
Laboratories do not directly publish software products that attract hundreds of millions of users and billions in revenue. Laboratories do not hire armies of lobbyists to control the regulation of their work. Laboratories do not compete for tens of billions in external investments or announce many-billion-dollar capital expenditures in partnership with governments both foreign and domestic.
But I still hear it all the time. We call them labs, even when they’re ready to IPO for hundreds of billions of dollars or pouring money into congressional races.
Let me add a few points.
First: They were never labs.
It’s not that these were harmless scientific laboratories that mutated into dangerous corporations. Any private AI lab trying to compete to build frontier models was always an AI company, or it was destined to fail.
That’s because there was no way to make the math work otherwise. The investment that it takes to train competitive frontier models—the talent, the infrastructure, the compute, the lawsuits—is in the hundreds of billions of dollars per company. Where was that money coming from? Three possible sources:
- Government
- Philanthropy
- Private capital
Government does provide grants, but nothing near this scale—this is closer to the scale of the Defense budget than scientific grants.
Philanthropy isn’t going to provide the money. Aside from the poor fit of frontier model development with philanthropic commitments, the scale is even further off. Even a consortium of the wealthiest people on the planet would drain a proportion of their resources to fund even one company.
That leaves private capital, and that comes with the expectation of enormous returns. All of the profit-maximizing behavior we see today was inevitable from the beginning, whatever games with corporate structure we played at the beginning. Everyone involved in founding these companies is smart enough that they knew, or absolutely should have known, that this is where we were headed.
This is not a story about how labs became companies. It’s about how they ever convinced anyone they were anything but companies.
Second: There are no labs inside companies.
The post I linked above says:
To be clear, there are labs inside many AI companies, especially the big ones mentioned above. There are groups of researchers doing research at the cutting edge of various fields of knowledge, in AI capabilities, safety, governance, etc. Many individuals (perhaps some readers of this very post!) would be correct in saying they work at a lab inside a frontier AI company.
Here I disagree. If you work inside a company and use its resources, your work is subject to its incentives and control. For a time, you may feel like an independent scientist with academic freedom. But if your work could accelerate (or hinder) the company’s competitiveness, that firewall will come down. If it doesn’t, you’re not really “inside” the company—you’re that company’s own philanthropic project.
Third: This is what science looks like
The first two points probably read as critical of AI companies, and they are—not because they’re for-profit companies, which is the only sensible approach, but because they pretend otherwise.
This final point is more of a defense. One reason companies like to call themselves labs is to appear less threatening. Implied is the idea that science is non-threatening, that we can advance the frontier of human knowledge without hurting anyone. I think about the “In this house, we believe” signs that include “science is real” and, just below that, “love is love” and “kindness is everything.”
This is a view of science defanged. The pursuit of scientific understanding has often been disruptive and destructive, tearing apart the natural world, human communities, and even our understanding of our place in the world. Scientific research is intertwined with the engineering that puts it into practice (and creates the pull for further discovery), as well as the quest for profit and power that underwrites it.
I don’t want to call these companies labs, because that word leads us to inaccurate expectations about their behavior. But I also want us to stop thinking of labs, and science in general, as harmless. The race to build more capable models and eventually create true intelligence is science. In fact, it’s arguably the most important scientific undertaking currently underway, and it can’t be carved off from the broader project of attempting to understand the universe. The fact that it is bound up in greed and power and violence is par for the course. We should be clear-eyed about that.