The (AI) Axeman Cometh
The following Guardian article makes some of the salient points that I think we should be taking notice of:
The AI industrial revolution puts middle-class workers under threat this time
Larry Elliott
https://www.theguardian.com/technology/2023/feb/18/the-ai-industrial-revolution-puts-middle-class-workers-under-threat-this-time
-
- AI in the form of generative LLMs will take middle-class jobs, unlike in previous automation revolutions, which largely took manual or highly repetitive jobs.
- automation both in the past and now has always tended to increase inequality by pushing money to capital owners and pushing manual workers either into destitution or into worse paid, lower quality work than they previously had; the middle classes are about to get the same treatment and I doubt they will like it.
- silicon valley is keen on UBI.
On point 3, I would point out that those in power say it cannot be paid for and will stop it from happening: expect widespread strife, misery, and in some places violence and revolution.
The following is also interesting:
From retail to transport: how AI is changing every corner of the economy
artificial intelligence in everyday life
Artificial intelligence has implication across the board, solving problems and raising others
AI industrial revolution puts middle-class workers under threat
Joanna Partridge, Phillip Inman, Alex Lawson, Jasper Jolly, Richard Partington, Gwyn Topham, Kalyeena Makortoff and Sarah Butler
Sat 18 Feb 2023
https://www.theguardian.com/technology/2023/feb/18/from-retail-to-transport-how-ai-is-changing-every-corner-of-the-economy
However, the authors do not mention the field of software, which could undergo the biggest revolution as software is something that generative AI is getting better and better at writing. It is also getting better at detecting errors in software; lots of coders are now using AI to debug their code or debug code written by AI. I think this will get us to an early version of Kurzweil’s singularity perhaps as soon as this year*, ie AI reliably able to write and test software that it then bootstraps to make itself better.
There are already AI algorithms that can learn (unsupervised, this is the key) to do highly complex tasks in a virtual world, such as play football where they know the rules or achieve tasks without knowing the rules, just being told what the goal is. At first these algos took years to learn (computer years that is, but in real time, just a few weeks); however, now, for some tasks, they can learn in seconds. These advances have taken place in less than a year. As I have said before, having watched this for a long time, advances which used to take years are now taking months or even weeks.
* Ray Kurzweil predicted the singularity by around 2045, but he seemed to think it would need AI to surpass human intelligence, which is generally taken to mean AGI or artificial general intelligence. I think Kurzweil is about to be proven wrong: you don’t need AGI, you just need something that is good enough to pass as AGI, ie Searle’s Chinese Room. It doesn’t matter that the computer actually understands nothing, at least in a human sense. This is just irrelevant as long as humans accept what it says as being intelligent and follow its instructions – which increasingly they do, as most people are now little more than clones, consumatrons and corporate robots.
Fintech is mentioned, but I wonder what AI will do more widely to the financialisation revolution that is so much a part of the neoliberal destruction machine? Will it make it go even faster or might it start, accidentally, to throw a spanner in the works?
The same question might be asked of the large consulting businesses, which are effectively stormtroopers for the same system of destruction.
Finally in this brief look at how AI will change much of human life, we have yet to see what AI-boostrapped software can do for robotics. Training digital twins is already amazing: I have seen examples where a physical robot starts off already knowing how to operate in the real world because it has been pre-trained in a virtual world. It is like being born at the mental age of twenty, having been able skip the previous two decades of painful and tedious learning.
Learning is actually what humans should be glorying in, as it is one of the things that the human brain was most ‘designed’ or rather evolved to do, but now learning is a horrible, miserable, anxiety-inducing wretched torment for vast swathes of young people. Indeed, increasingly, for all those of us who actually want to think, learn and do something other than what the mainstream consumer machine wants us to, we are punished horribly by this system.
If, as is surely the case, the new situation demands different ways of thinking, we can be fairly sure that most of that thinking won’t be forthcoming, because the universal group-think of neoliberalism and libertarianism has reduced most people to being little more than Manchurian candidates. The irony is that most people are already robots: the big question is how intelligent and autonomous they really are. Maybe machine robots will have to teach humans how to be human?