A Mathematician, a Different Kind of Mathematician, and a Statistician
PhD, public policy | AI & ML | Too many affiliations to list | he/him/Dr
| Website | https://jameshoward.us |
| Github | https://github.com/k3jph |
| https://www.linkedin.com/in/k3jph/ |
Someone complained their degree didn't get them a job. The response: what was your major? Both refrains. Both miss the point.
I have an area studies degree and I'm making espresso. Got it in 2020, already a working data scientist, because I wanted to know things. The espresso is just coffee.
College is not job training. It makes you a better person.
Researchers who had 'artificial intelligence' on their CVs in the late 1980s quietly rebranded as 'machine learning' practitioners. That wasn't cynical. It was accurate. 'Machine learning' described what the surviving methods were actually doing. The AI winter sorted the field. Those who stayed built what came next.
https://jameshoward.us/2026/06/04/the-second-winter/ #AIHistory #MachineLearning
MYCIN outperformed human specialists at diagnosing bacterial infections in 1979.
It was never deployed clinically. The liability question it raised has never been resolved.
New post on the expert systems era: https://jameshoward.us/2026/05/20/the-rule-based-interregnum/
The first AI winter wasn't a failure of intelligence. It was a failure of honesty about what the hard problems actually were. New post on the combinatorial explosion, the frame problem, and the Lighthill Report.
We are at the 1995 moment for AI governance. The principle: AI systems making consequential decisions about people should reflect the values of the jurisdiction where those people live. Not the values of whoever built the system.
The trolley problem has a nationality. The rules for the machine should too.
https://jameshoward.us/2026/05/11/who-makes-the-rules-for-the-machine/
#artificialintelligence #AIgovernance #algorithmicsovereignty #publicpolicy #GDPR
