- by RJ Harris
- The Guardian
- Issue #2200
Photo: www.vpnsrus.com – wikimedia.org (CC BY 2.0)
As a worker are you worried about “AI” taking your job? Or are you worried about your boss drastically reducing your work hours after you teach an algorithm everything about your job?
Currently “artificial intelligence” (AI) tools are underpriced. This is to attract a customer base. I argue that the priority customer base is institutional rather than retail.
Retail customers – individuals like you and me – do not generate the kind of inertia which will keep them tethered to an AI solution while the price of that solution increases exponentially. Institutional customers must develop processes and procedures around the adoption of new technologies and stick to these technologies as the price rises.
A thought experiment: A state government, under budgetary pressure, introduces under-priced AI tools to streamline clerical processes. This allows them to increase the intensity of work for their human workers. Workers who struggle against this technological adoption are fired through whatever mechanisms are available – clause 11 restructures, “realignments” and so on.
After twelve months the remaining workers have little trouble with the AI workflows. They also understand very little of the old backend processes which the AI workflows now streamline. That’s the stuff you don’t see, but which makes what you do see possible.
The departed AI-oppositionists comprise the people who understood the workflow from beginning to end. They understand the work from the generation of some kind of datapoint in the community, its communication to the workplace, the manipulation of the data in the workplace, the entry of a datapoint in various databases and the interpretation of this new entry in larger datasets. They’re gone.
After a period of time, 12 or 24 months or so, the company which owns the AI comes to the state government department with a new cost for their entrenched tool at an increase of 1,000%. The government institution is suddenly under massive pressure to pay this new rate as they have lost their institutional knowledge of the old system processes in favour of a workforce who understands only AI-managed workflow. In an effort to extricate themselves from the class-struggle of a workforce which demands a higher rate of pay per cycle and will strike to get it, they are facing unaccountable and giant multinational companies who can reduce the operating capacity of said department to zero at the end of a billing cycle.
The class struggle is being shifted from proletarian-bourgeois to intra-bourgeois conflicts. Imagine your boss, Serco, in negotiations with Anthropic over the cost of Claude Enterprise. In this scenario Serco is facing a 1,000% cost increase. Enter you: you want a 10% increase to your pay.
It’s you, the worker – already viewed by the company as stupid meat – versus Anthropic, viewed by the company as an office of seers who produce this amazing black-box product which works magic. In this scenario, Anthropic is indispensable, and you are all too dispensable.
If you thought that hiring 22-year-olds at KPMG to tell your company to increase productivity by firing staff and raising prices was stupid, just wait until you get the sack because a 25-year-old (the same person just three years older) has figured out that the executives want to hear nothing more than “MORE AI” and they are happy to provide.
For those of us who are in manual work this scenario is only so scary. There is so far close to zero application for Deepmind to dig a hole or ChatGPT to wipe someone’s bum. However, in the office not far from a bum-wiper sits a clerical worker under the pressures described here.
If AI is to be used in our workplaces it must be in our service – not a multinational and not the boss. In service of: reducing our work intensity; increasing workplace safety; increasing our share of productivity and; securing our jobs for the future without destroying the planet.
Part of our struggle is communicating this to our workmates. The thought experiment described here has not yet happened, but without organised struggle this technology will be used to increase the intensity of work and undermine our job security.
