AI Identity Crises
The Problem is Epistemic. The Solution is Not. Samuel Bagg for APA Online, 12 Nov. 2025.
Knowledge formation is driven by social identity. So political identity changes what you think is real. I imagine most people realize this by now; the research in the article linked above doesn’t surprise me in the least.
I had a personal revelation on this as well. lots of people are mourning the craft of software, the identity of being a ‘software engineer.’ In short, AI has made it all to obvious that the California ideology was always a mistruth.
The death of software artisanship is accompanied by the realization that the yuppies who recruited the hapless hippies never really valued their expertise; it’s a (re)awakening to the hungry realities of capital and its voracious appetite for exploitation.
Argument: By outsourcing the very intellectual labor that built Silicon Valley, the turn toward probabilistic models has proven the contradictions inherent in the Californian Ideology.
The Californian Ideology
- Yuppies and Hippies - the contradictory marriage of free spirited culture and free market policy.
Automation
- The Californian Ideology relies on a notion of software artisanship. The software developer, as a craftsperson, purportedly has a unique perspective on the craft. Much like the automation of skilled labor in the first industrial revolution left many skilled craftspeople resentful and without work, the current shift toward probabilistic models and the consequent laying-off of thousands of software developers, artists, and writers has lead to a wave of anti-AI sentiment. There has not been any violence against datacenters - yet.
Identity and Ideology
- Identity and ideology are intrinsically linked. Many developers have found themselves in crisis of consciences, or existentially fraught.
- Ethical concerns surrounding the training and development of LLMs.
- Existential questions surrounding identity.
- Transformer models in their current iteration are associated with the political far-right due to the exploitative nature of their training and their widespread use among the the far-right.
3 claims:
the tools have changed but the craft hasn’t
- what else is a craft besides the particular quality of labor involved in the creation of an artifact?
- software craftsmanship movement in early 2000s already saw a big shift and attempt to formalize the craft of software. vibecoding is very different.
ai has completely changed software
- not entirely true, but not wrong. many changes are social, like the shift from fully open to semi open code bases (the sqlite model)
- traditional coding is still possible but no longer the business requirement. high velocity slop is the norm. it will crash and burn and we’ll see more emphasis on human expertise
nothing is different
- only true if you have your head under a rock. programming is social, it requires cooperation, and llms are changing how that works.
- people don’t want their code scraped, so they hide it, reducing the capitalist incentive to steal and contribute nothing