Over the last 200 years, society became steadily more liberal. Serfdom and slavery were largely abolished, democracy exploded and most women and children are free from violence. Redistribution is widely practiced and the median person is far richer, healthier and freer than ever before.

We see this progress at the generational level: people are less racist than their parents. We also see it on the historical scale: in 1512, public torture, marital rape and slavery were acceptable and practiced widely, now they are largely morally repulsive and very rare. Peter Singer calls this phonemana: “Moral circle expansion”. MLK said “the arc of the moral universe is long, but it bends toward justice”, and Moldbug cynically wrote: “Cthulhu may swim slowly. But he only swims left.”

Of course there were setbacks along the way, like the Armenian genocide, the Stalinist purges or the Maoist cultural revolution, but in general, progress is strong and liberalism is winning. The rate of “liberal progress” might even be accelerating.

Why is that?

Why is society getting more liberal? There are many explanations. I find the “mainstream economics” one to be by far the most compelling. It goes something like this:

Institutional Selection

As the economy gets more sophisticated and dependent on human capital, “being good” becomes more advantageous. Universalist moral codes, impersonal trust, and individual rights produce cultures that cooperate more effectively at scale. States that adopt liberal institutions tend to prosper and dominate.

Free people are more productive than slaves and serfs. Free speech leads to better ideas. Free markets and fair rule of law lead to innovation and economic growth. Happy and engaged masses win wars. Women rights are good for productivity. Happy and educated kids become more productive adults. Fair international trade is better than extractive conquest. Economic mobility and meritocracy improve governance.

This is the story of the 19th century (aka Pax Britannica) and the 20th century (aka Pax Americana). The last two centuries have many cases where local elites push through liberalization reforms after getting their asses kicked by the West. Most famously: the Meiji restoration, the Perestroika and Atatürk’s reforms.

Not all countries are liberal democracies. Why is that? Economists talk about the “resource curse”: resource rich countries are typically authoritarian oligarchies. If the elites get their power and money from resources, they don’t need to “invest” in their people or infrastructure. Poor, uneducated subsistence farmers are easier to control! Elites in such countries get stuck in a local minimum: investing in the populace has a long and uncertain “payback period” and might backfire, oppressing the people and extracting minerals has immediate ROI. People and elites in liberal democracies are mostly fine with this. I think that’s the strongest argument for the institutional selection^[1] view

Will the trend continue?

Why not?

AI!

Robots and AIs might displace the average human from the labor market. Drone armies will make human soldiers and policemen unnecessary. The value of the average human to elites will diminish significantly. As a result, treating the average human nicely and fairly becomes less important to the survival and flourishing of the state and the average person will be gradually or quickly disempowered^[2]

Why yes?

There are many cases where someone is less empowered but society still treats them with respect. Some examples: indigenous groups that maintain special rights (Native American reservations, Hawaiian lands, Maori lands), affirmative action recipients, protected species and very old people.

The strongest example is countries who are “resource economies” but still have inclusive institutions. e.g Norway, NZ, Iceland, Canada, Australia and Botswana. Amazingly, there are basically no examples of countries that became less liberal after “striking gold” and finding a bunch of resources.

Obviously there are cynical “realpolitik” explanations for all of the above, but I’d like to think that inclusive institutions are just very robust. Once you “include a group”, it’s very hard to exclude them again, maybe for collective action or “slippery slope” reasons:

“First they came for the average person
And I did not speak out
Because I was not the average person

Then they came for me
And there was no one left
To speak out for me”

Perhaps this will stay true forever? I hope so.

What can we do now?

While we (non-elites) are still empowered, we can pressure the institutions to change in a way that will make society more robust to disempowerment. Some quick ideas:^[3]

Transparency

We could make our institutions and elites more transparent and legible to the average person. This might take the form of frequent public audits of AI companies and the regulators.

There should be strong incentives for sophisticated actors to investigate companies for violations / increasing x-risk and report those violations. We can use existing mechanisms for that, e.g short seller firms, class action litigations, market competitions (“my competitor is breaking the law!!”), or new mechanisms like generous whistleblower programs for AI insiders. Better epistemic tech would help the masses understand their institutions better and take better advantage of transparency.

Decentralization

We can make it harder for elites to coordinate. AI development should be distributed across multiple companies both vertically and horizontally. Ideally, those power centers are somewhat adversarial to each other. This way, key pieces of infrastructure aren’t controlled by a small group of people.

We should also encourage “AI insiders” to become “outsiders” via term limits (with long / infinite cooldown periods) for executives, top researchers, influential bureaucrats and power brokers. We could also have mandatory gardening leaves for AI executives to make “coordination by osmosis” less likely.

Ideally, AI capabilities are also distributed across multiple liberal democracies with international agreements. Something like: “we commit that we will stop trading with you and maybe even declare war on you if you wirehead your masses or become an autocracy“

Control

AI “constitutions” and AI values should be visible to and influenced by democratic institutions, ideally in ways that would “lock” those values in. Perhaps governments establish agencies that audit training procedures and evaluate models to make sure that they don’t concentrate power and are aligned with the needs of the average person. AI constitutional amendments could require a supermajority vote like US federal constitutional amendments. We can go even farther and demand that AI values can only be changed with some sort of international agreement between liberal democracies.

Differential technological development

We should slow down and limit technologies that disproportionately help the elites gain power: mass surveillance, AI propaganda, autonomous weapons and autonomous policing.

^
Another good one: there is a lot of redistribution within liberal democracies, but there is almost no global redistribution from rich countries to poor countries. Why is that? International redistribution doesn’t have strong institutional selection for it, but national redistribution does
^
Disempowerment isn’t even the worst case scenario. An AI takeover can lead to total human extinction or even something far worse
^
I found these articles inspiring: breaking the intelligence curse, mitigating ai enabled coups, mitigating gradual disempowerment

Is Progress Inevitable?