May 19th, 2023
by Andy Didorosi
Freshly stirring the AI-taking-our-jobs pot is British telecommunications monopoly BT bragging about how it’s laying off thousands of workers due to new AI efficiencies over the next decade. As recently as a few weeks ago, corporations used to couch the term “layoff” with feigned deep regret and remorse as something “they’d prefer to not do but their hand was forced by the economy, aw shucks.”
Now, with BT’s stock price in the tank, they’re promising how hip and futuristic they’ll be very, very soon with cheap, profitable AI agents telling you your bill went up instead of expensive, fussy real humans.
But the question keeps popping up in my mind: why now? What precursor now exists making “AI” the thing instead of any number of other similar “job-destroying” technologies?
The industrial revolution holds the key. Namely textile machines, and then COVID. I’ll keep it short, this isn’t history class:
It’s a common fact that industrialization started in Britain – big machines belching black smoke turning out cheaper products than their handmade competition. During this era in Britain human labor was very expensive yet energy, namely coal, was very cheap. Not so much in London, but in nearby Newcastle where the coal mines were located. Out of these conditions came the steam engine. First a pretty bad one that used tons of coal to put out very little work, but over time they got better with greater efficiencies and more work they could do. First pumping water, then spinning textile mills and ship propellers.
But why Britain? With so many other places so far behind Britain in industrialization, some nearly 100 years, it must just be that Britain had a special pluck and engineering culture to get this done, right?
Wrong. The other major players in the early 1700s including France, China, and the Netherlands all had similar cultures of innovation and creation as Britain did. The major difference is Britain had both extremely expensive labor and cheap energy. China and France still had relatively cheap labor in the early 1700s, but quite expensive energy costs. It was just cheaper in these places to make things by hand. The invention and implementation of the steam engine wouldn’t have saved China or France any money – it actually would’ve cost them substantially more to operate in the near term.
The magic formula is this: Innovation happens only where it becomes cheaper after both development costs and implementation.
This is the path that all industrialization technologies followed through the 1700s and 1800s and not just relative to energy or labor. Mechanized textile mills preceded the steam engine so energy prices weren’t as relevant in this case, but these had a different gating resource: gears.
Automated textile mills like spinning jennys were relatively expensive to build, requiring specialized parts from the watchmaking industry. These gears weren’t too difficult to access but cost substantially no matter where you were located. The price of the machine, driven by the cost of gears, only made the development of the machines possible in places where labor costs were sufficiently high.
Innovation requires investment, and investment only happens on a positive expected return.
That’s where we join back up with AI, machine learning, and BT bragging about layoffs. The AI boom is only made feasible now by labor costs skyrocketing through COVID making AI the cheaper option. Workers received both healthy unemployment benefits plus a heaping dose of reality about the brevity of life during the immediate CVOID crisis — which put big upward pressure on wages. Training large language models (LLMs) and developing AI technology is just expensive. This development wasn’t economically feasible before when worldwide labor was cheap, but the combination of a higher labor-to-energy cost ratio and proliferation of cheap compute time means it’s AI time.
There are plenty of other factors to be sure, but one somehow always reigns above all: profit.
Elsewhere on the internets:
Where have all the hackers gone? https://morepablo.com/2023/05/where-have-all-the-hackers-gone.html
My 20 year career is technical debt or deprecated: https://blog.visionarycto.com/p/my-20-year-career-is-technical-debt
The simple joys of scaling up: https://motherduck.com/blog/the-simple-joys-of-scaling-up/
Why is my dryer radioactive? https://physics.stackexchange.com/questions/764460/why-is-my-dryer-radioactive
Quid Pro Crow: https://languagelog.ldc.upenn.edu/nll/?p=58724
My Inventions - Nikola Tesla: https://archive.org/details/MyInventionsNikolaTesla/page/n11/mode/2up
User-Driven UI: https://garden.bradwoods.io/notes/design/user-driven-ui