Lots of writing about the future takes one of two forms:
Dystopian: everything is awful! Robots/Fascists/Communists/[Insert baddies here] have brought about the apocalypse, or made most people into their prisoners or slaves.
Utopian: everything is amazing! Robots/Communists/Philosophers/[Insert goodies here] have fixed the world, and now there’s a new type of Femo-anarcho-libertarian-socialist-capitalist-democracy where everybody votes for and experiences good things and doesn’t vote for or experience bad things.
But unless we get the 5% chance world where there’s an existential catastrophe by 2072, or the 0.1% chance world where AI somehow fixes every problem known to man (feel free to contest the percentages, I didn’t think that hard about them), the most likely option is that most things get a bit better, some things get a lot better, and a few things get a bit worse, right? So, here’s my fictional account of the ‘boring timeline’, where things go mostly pretty well but some things also go pretty badly.
Although we never got the heralded Superintelligence, predictive algorithms became extremely good. In 2040, some developers introduced a few adjustments to the latest release of XGBoost so that it could make an assessment about whether someone being tried was guilty or innocent. When it was tested on previous known wrongful convictions, the algorithm determined that the defendant was innocent about 85% of the time, and as the algorithm was improved over time, it became clear that its assessments were much better than those of humans serving on juries.
The algorithm was introduced into the American criminal justice system extremely gradually. At first, the algorithm was only used for people who were appealing their convictions - if you were found guilty by a jury, you could have the verdict examined by the algorithm. If the algorithm determined that there was a high chance you were innocent, you would almost definitely be freed. Eventually though, states started simply replacing human juries with the algorithm. The Supreme Court determined that this was not a violation of the Sixth Amendment, as the algorithm was less biased than human jurors.
There were definitely problems - one journal article showed that ethnic minorities were much more likely to be found guilty by the algorithm than were white people, although a prominent academic pointed out that this was even more true of human juries. There were also questions about what ‘reasonable doubt’ meant - because the algorithm actually gave a percentage chance that a defendant was guilty, a decision had to be made about the threshold for which the defendant would be sent to prison. It was eventually decided that the algorithm’s determination had to be that the chance that a defendant was guilty must be over 97% in order to be found guilty. Many opinion pieces were written.
A few weird things happened in the past few decades. Right-wing populist parties started to dominate European Parliament elections. Obviously there had been signs of this earlier, with UKIP and subsequently the Brexit Party doing extremely well in European Parliament elections before the UK left the EU, and the Rassemblement National winning the EP elections in France in 2019. But that was just the beginning - in 2043, Identity and Democracy merged with the European Conservatives and Reformists to form Europeans for Radical Democratic Reform, and they became the largest grouping in the European Parliament in 2053. But this success in Europe did not translate into domestic success, leading to a rather odd situation wherein EU legislation became more right-wing than domestic legislation in most countries (although freedom of movement remained in place), and support for the EU fell dramatically. People on the left became much more likely to say they supported their country leaving the EU, but there was no corresponding increase in support for the EU among supporters of right-wing populist parties.
Neo-Luddite parties emerged in several countries in response to the rise of Artificial Intelligence. Because AI was mostly replacing the jobs of writers, artists, programmers, lawyers, and other white-collar workers, these parties are largely supported by people who would otherwise be voting for centre-left and centre-right parties. In the UK, which fell behind Italy and Spain in GDP Per Capita in the mid-2040s, the Labour Party went through an ‘if you can’t beat ‘em, join ‘em’ moment, and decided to go full Neo-Luddite. In a proposal widely regarded as unworkable, they campaigned for a Robot Tax that apparently would create 600,000 new jobs for human beings. They lost the subsequent election to the Conservatives. Many opinion pieces were written.
Genetic Engineering is pretty popular now. Most people agreed that getting rid of sickle cell anaemia and other debilitating diseases was a good idea, and sickle cell disease was eliminated by 2045. The IQ stuff was obviously way more controversial, but once it started happening in China, the US quickly followed suit. The favoured technique is polygenic embryo selection - embryos are selected on the basis of their ‘polygenic score’, a score for an embryo’s genome that is correlated with IQ. Parents then choose the embryo with the highest score, and see a moderate IQ bump in their child. The mean IQ for a child born in the United States today is 25 points higher than a child born in 2022.
People had originally assumed that the inequality created by genetic engineering would take place within developed countries - the super-rich in the United States would have babies with insanely high IQs, and the wealth gap between the richest and poorest Americans would quickly increase. In fact, the main inequality created by genetic engineering was inequality between countries. Developed countries were pretty quick to realise that offering free polygenic embryo selection to all parents was the only way not to fall behind, so that happened in all countries that could afford it. Developed countries did eventually start paying for polygenic embryonic selection in countries that couldn’t afford it, but genetic engineering in Subsaharan Africa and poorer Asian countries happened much slower than in the developed world. Many opinion pieces were written.
Every country is a lot richer than it was in 2022, although there’s definitely some (mildly depressing) counterfactual world in which every country is richer still. GDP Per capita in the United States is around $180,000 in 2022 dollars, and the equivalent figure for Subsaharan Africa is around $26,000. Let’s run through the things that weren’t the cause of everyone being a lot richer:
Voters didn’t suddenly and unexpectedly opt for better economic policies - no country went full open borders or anything like that, although in the 2030s a few countries did join the EU.
Institutions in developing countries didn’t suddenly become much more inclusive, although most countries became gradually more liberal and democratic over time (apologise to Fukuyama when you get the chance).
It wasn’t the boost in IQ that led to huge economic growth. Growth was relatively stable in most developed countries, there wasn’t some growth explosion after genetic engineering became popular, although growth has been slightly faster since parents started tinkering with their childrens’ cognitive abilities.
There wasn’t some new technology that made everything just get a lot better - Artificial Intelligence has had some impact on growth and has replaced the jobs of some humans, but it didn’t lead to a quadrupling of world GDP or anything totally nuts (or, conversely, a mass unemployment situation).
So, what was it? The answer is the boring one. Things gradually got better over time through new technologies (not one single earth-shattering technology) and innovations, and more people are well-educated enough to be very productive in the world economy. That’s it, the secret sauce is just education and compounding returns over time. Again, many opinion pieces were written.
This article was written as a submission to the Blog Prize’s May mini-prize. Go and check out the Blog Prize and start a blog if you haven’t already!
I think it is significantly more likely that it would be considered a violation of the 6th Amendment to use them as a substitute for a jury. Originalist thoughts will disagree with it simply because it is new and goes against the history of trial by jury, and liberal Justices will be skeptical for the biased reasons you mentioned. I do suspect the algorithm will be admitted as evidence at one point, but even that will require the supreme court to either overrule or make a large carve out from People v. Collins. There is still a large chance that algorithms will be treated the same as the evidence of statistical improbability in that case, and be completely banned.
Edit: People v. Collins was a California supreme court decision, so technically it only applies in California, but most states treat it similarly.
All pretty plausible. Though I think in this world, gains would indeed be driven by IQ and AI.
Also I want an apology from Fukuyama. “End of history” lol.