The Enemies of Progress
The Progress Movement Can Never Win Unless We Understand Why We Lost
Time and again, we’ve seen emerging scientific fields fall to stagnation. We’ve seen it with nuclear, pharmaceutical discoveries, and high-speed transport. If not for operation warp speed, it would have happened to mRNA, leading to hundreds of thousands more COVID-19 deaths. Regulation on the inputs and outputs of emerging technologies at scale are a spider’s web, laying in wait for new entrepreneurs and scientists.
I’ve read variations on this story countless times:
Scientists struggle for decades to make a new scientific breakthrough. They discover a widely applicable use of their research that signals towards a bountiful field of future research. They see the good it can do for ordinary people and want to make that happen. They try to commercialize their research. Then they find out that all sorts of bottlenecks that were put into place to restrict other fields now matter to them. Both the inputs, fundamental ones, like energy or talent, as well as the outputs, commercialization, diffusion, operating as a business are tightly constrained. The disappointment affects the entire community, field of research, and scientific endeavor.
I spoke to a founder yesterday that remarked that the exact same thing might just be happening now with AI. The theory of inevitability, that artificial general intelligence, an AI model capable of doing every human digital task, will just happen, will just materialize. It doesn’t require more fundamental research, it doesn’t require more difficult execution. It has led the very researchers who we would need to accomplish such a fanciful task to abandon that fundamental research. It has led some of them to go on a mad dash to capitalize, to make a ton of money for themselves while they still can. And it has led a dangerous fringe into advocating for slowing or stopping AI research, which we all know from the historical precedent, would cause a deadly stagnation killing hundreds of thousands, if not millions of Americans.
This is what Peter Thiel calls indefinite optimism. Indefinite optimism is a kind of ostrich behavior — sticking your head into the ground and ignoring all of the challenges up ahead.
Instead of believing in a concrete progress in which you look into exactly what technologies will build the future, exactly what discoveries need to be made, exactly what bottlenecks in policy are stopping those technologies from becoming a reality. When you agglomerate all of those things into an inevitable prophecy, it makes you lazy.
The other failure of indefinite optimism is complacency about government. When you ignore the consequential policy that has happened in the past and will happen with the politics of the present and future, you make the spider’s web which has sabotaged past scientists and founders inevitable.
I’m going to make a statement that is both trivially and highly controversial — there are enemies of progress.
There are people who if they had not existed in the past would result in a vastly more competent, more advanced, and more healthy society today. It includes zealots who arrested and censored Galileo for daring to suggest that the earth revolves around the Sun. It involves those who protested nuclear energy and resulted in the establishment of the Nuclear Regulatory Commission, an agency that has unilaterally banned new nuclear power plants for 50 years. It includes those who cut off access to life-saving drugs for bureaucratic compliance, leading to the “invisible graveyard”. They are responsible for the deaths of hundreds of thousands of Americans.
If you believe that progress is a fundamental moral good, then you must believe that there are people who are a fundamental moral evil.
Right now, the Spider’s web of legacy regulation is tangling up AI. Energy permitting laws make it far harder to scale up the energy requirements necessary to train the next generation of models. Disparate impact law narrows the window of allowable opinion, restricting the ability for AI to make factually correct statistical observations with possible downstream effects on overall performance. A highly volatile and arbitrary copyright environment makes it so that whether we’ll have enough data to train the next generation of models might be decided by a few judges.
Startups require a singular focus. Each component that takes resources away from the central research, the central product, and the central distribution of a startup splits that focus and splits that progress into exponentially less efficient pieces. All of this is to say that removing current bottlenecks on AI are a massive public good, even to the most competent of AI researchers and founders. Your decisions matter.
The other side of the coin is that history repeats itself. The past saw false and baseless moral panics around factory machines, nuclear energy, and GMOs. The past saw religious zealots attempting to censor and control ‘dangerous’ information. The universal pattern in human psychology is that if you let your imagination run wild, it will result in fear, panic, and authoritarian control, regardless of the underlying technology. That instinct has been disastrous for our technologies. It has been disastrous for clean energy, for medicine, for housing, and for transportation.
We cannot let well-funded activists with a pseudo-religious fervor destroy a burgeoning scientific field based on purely hypothetical doomsday scenarios based on no empirical evidence and understandings of a technology that are long outdated. I am of course talking about the anti-nuclear movement, which used fears of outmoded reactor models and apocalyptic rhetoric to pass the Energy Reorganization Act of 1974. These nuclear Doomers claimed that the Nuclear Regulatory Commission would provide sensible guardrails on nuclear energy, when it reality was as a de-facto ban on nuclear energy for 50 years.
Definite optimism requires a clear view of the underlying technology. That is obvious. What isn’t obvious is that definite optimism requires a clear view of politics. The thousands of nuclear physicists failed to defend their industry in their seventies. It rarely matters how brilliant your research is if it is illegal.
Defending definite optimism requires admitting technology has friends and enemies. Time and time throughout history, there are people who refuse to be convinced by any amount of empirical evidence. Their goals are fundamentally antithetical to technological progress. And the only way to defend our future is to reduce the impact they have on government and society. ☐
Somebody should write a book...
https://amzn.to/3z4lDKj
>And it has led a dangerous fringe into advocating for slowing or stopping AI research
The "dangerous fringe" including the very set of pioneering AI pundits, AI researchers and AI enterpreneurs, who have all warned about its dangers, sometimes calling them "existential"?
Right or wrong, those are not some crazy fringes, they're at the very core of AI. The rest is their employees, marketeers, and resellers.
> which we all know from the historical precedent, would cause a deadly stagnation killing hundreds of thousands, if not millions of Americans.
If "slowing or stopping AI research" is enough to cause "a deadly stagnation killing hundreds of thousands, if not millions of Americans", then perhaps the country is in dire need of a balanced robust multi-faceted economy - and has given on that, hoping for miraculus technological solutions with religious fervor instead.