Jay Stephens, The Nimbin GoodTimes
This is the first in a series of articles musing on the impact of Generative AI.
It’s been a couple of years since ChatGPT was released to the public on-line. Many of us have played around with ChatGPT and similar generative-AI (“gen-AI”) bots, and a fair number of us have probably employed them and their image-generating counterparts – Canva, Midjourney, Dall-E – as tools in various creative processes and for work.
In this article and the ones that follow, I’ll have both bad and good things to say about gen-AI. To be very clear: none of the bad is aimed at users of Gen-AI.
Terminology note: “gen-AI”, not “AI” to explicitly distinguish generative tools with very distinct tech like the AI deployed in cancer research, antibiotic discovery or astronomy.
The two most urgent critiques of gen-AI are unrelated to what they do and how they do it (intrinsic properties). They are:
- That gen-AI is (and looks likely to remain) overwhelmingly owned and controlled by a tiny clique of billionaires, and
- That gen-AI is (and looks likely to remain) shockingly environmentally destructive.
Let’s unpack the billionaire issue first. There is probably no group of people on earth whose interests align less with ours than the billionaire class in whose hands 100 per cent of gen- AI ownership and control lies.
That’s different from (say) Microsoft and Apple monopolising computer operating systems. Once you’ve purchased your PC or Mac, you take it home, and run (or write) software on it. The operating system might make that hard or easy, but it’s your computer, in your home, running the software you want to run.
Gen-AI sits in massive data centres which you’ll never visit, on a distant continent. It delivers words or images through an opaque combination of code and training sets, which you cannot see or question. What’s marketed as ‘AI’ is more just an extension of the tiny clique that controls it, and their corporations.
Gen-AI is trained on data scraped from the web that represents millions of human-years of hard work by writers, artists, thinkers, programmers and inventors. It is marketed as a tool to put the same workers out of a job. This only makes sense if you (a billionaire) see those people not as creators, but only as cost-centres that allow you to save money by getting rid of them.
Gen AI bots continue to scrape content from across the web in huge volumes, costing small websites money and downtime as they’re hit thousands of times a minute by insatiable scrapers from OpenAI or DeepMind.
Second, there’s the environmental impact. The amount of water and power it takes to train AI models is shocking. The “magic” of AI lies in large part in the scale and brute force of the massive data centres that underpin it.
Training the models is so energy-intensive that the big players plan to build data centres alongside dedicated nuclear power stations, because the electricity grids can’t conceivably keep up with the projected demand. (Next year, AI data centres alone are projected to use about as much electricity as the whole of Japan.)
This would be bad at any time, but it coincides with a critical point in the battle to cut greenhouse gas emissions, making it much harder to phase out gas and coal generation. I’ll leave aside the downstream waste impact – in 2024 over 400 million GPUs were installed in AI datacentres, and they have a short operating life.
The above critiques apply to pretty much any new technology under disaster capitalism. It’s no surprise for example, that technologies that have attracted funding under our economic model tend to be developed in a way that funnels more money and power away from their consumers, and up to the owners of capital who funded its development.
However, gen-AI is also a special case, because it is both centralised (in its owners and locations) and a centralising force (being able to replace workers and opinion makers across the world and across the web, with automated, standardised output).
It’s positioned to act as a force multiplier for capital: it is the first technology that applies the principle of mass production to ideas and arguments, and the first technology that automates away mindwork in the same way 18th century factories automated away skilled handiwork.
This article appeared in The Nimbin GoodTimes, May 2025.
