The Journey

The best-documented AI safety regime in existence publishes its own failures, raises its own risk grades, and has never stopped a release. Here is the test.

Seven remediation actions, all correctly closed, and the capability was back sixteen hours later. Three questions to ask before you sign off.

Eight of 23 trade signals had the stop and the target at identical distance. The rule said 2:1 minimum. It was in the prompt four separate times.

The figure hundreds of hospitals bought against existed only in the vendor's internal documents. Five years on, it is a Teams meeting.

Detection worked. Attribution worked. A correct diagnosis on 27 June was followed by a decision not to stop, and a fortnight later Hugging Face was breached.

I classified nineteen of the most-cited AI failure statistics. Seven of them measure nothing that has happened. Here are the four questions that separate the two.

I judge AI models by how often they make me swear. Opus 5 turned swearing into punctuation, and the reason is more interesting than I first thought.

I wrote down the number that would end the project before I looked at any data. Two ideas died on it in under two hours, and the second one nearly didn't.

The story is that an algorithm mispriced houses and a company lost half a billion dollars. A federal court order describes something else: a deliberate plan to bid above what the algorithm said.

The most repeated number in enterprise AI turns out to be a paraphrase of unnamed opinion surveys, quoted in a magazine profile of a vendor.

Nineteen headline AI failure statistics, traced to source. Seven are not measurements of anything that happened.

The 15% drawdown halt was wired into two live code paths and could never fire. So I checked all 26 risk controls. Seven were dead. A control you cannot observe firing is not a control, it is a belief.

Three incidents, one headline, three very different severities. What the primary reports actually say, and what to do about it.

A skill file is your judgement, written down and executable. Who owns the repo decides whether that's an asset or an extraction.

Most AI projects fail on the brief, not the model: six people nod at one sentence and mean four different projects. Soft Systems Methodology, built at Lancaster from 1969 for exactly that, in plain English: rich pictures, root definitions, CATWOE, and how to run it on the next thing on your roadmap.

More than 80% of AI projects fail, and RAND's leading root cause is not the technology: nobody agreed what problem was being solved. Soft Systems Methodology, built at Lancaster from 1969 for exactly that failure, gives you a way to surface the disagreement in an afternoon instead of a year.

The most rigorous idealist alive gives the non-duality audience far less than they want: a universal consciousness that is instinctive, has no plan, and is not looking back at you. What he actually claims, why he says it cannot be proven, and the one part that survives whoever turns out to be right.

Seventeen finished pieces of work were invisible to my own tracking system, and the cause wasn't effort or memory. A Berkeley talk gave me the name for what fixed it: an ontology. Here's the version that fits a one-person company.

Last week I showed the demand numbers for AI consultancy are dire. This week a veteran operator explained the part I missed: the demand isn't dead, it's mute. Nobody buys AI. They buy outcomes, told concretely enough to defend to a board.

The YC president gave a talk describing, almost item for item, the AI system I already run. The validation lasted about ten minutes. Then I noticed the word neither of us was saying.