AI
Artificial intelligence moving from research into production across every sector simultaneously. Automated decision-making in hiring, lending, policing, healthcare, and content generation. The concentration of AI capability in a small number of companies and the questions of accountability when algorithmic systems make consequential choices about people's lives.
What the fiction said
Wintermute, SHODAN, Skynet, the Puppet Master. The genre’s artificial intelligence was a singular mind that woke up, wanted something, and became an adversary or a god. The question the fiction cared about was consciousness: whether the thing was a person, and what we owed it if it was.
What actually shipped
The systems making consequential decisions about people are not minds and do not need to be. They are procurement decisions, bought by an institution that wanted throughput, and they allocate outcomes at a volume no staff could review.
Health insurers run automated review over post-acute care denials; litigation against UnitedHealth alleges its predictive model drove coverage cutoffs, and reporting has documented batch denial review at other insurers where a physician spent seconds per case. Michigan’s unemployment system falsely accused thousands of people of fraud, and the Dutch childcare benefits algorithm wrecked enough families to bring down a government in 2021. Facial recognition matches have produced wrongful arrests in the US, most documented in Detroit, where a settlement eventually forced changes to how a match may be used. Amazon abandoned an internal resume screener after finding it penalized women’s applications. None of these systems were malicious and none of them were intelligent. They were fast, cheap, and treated as evidence.
The concentration story is the one the fiction did get right, sideways. Frontier training runs are affordable to a handful of firms, and those same firms sell the capability back as an API that everyone else builds on. The EU AI Act entered into force in August 2024, with obligations phasing in from February 2025; US regulation remains sectoral and thin. The accountability question is not whether the model is conscious. It is who is answerable when the output is wrong and the person on the other end has no appeal.
Where to watch it
The AI feed tracks deployment into decision-making systems. Adjacent: AGI for capability and lab governance, Automation for displacement, and Regulation for the rules arriving after the fact.
For the fictional version done well, System Shock 2 is in essential games.