AI & Humanity · 6 min read
Could AI Become Better at Decision-Making Than Humans?
In narrow, well-measured domains AI already outperforms people; the open question is whether accuracy on a defined target is the same thing as good judgement.
ReadCategory
Essays connecting fiction, technology, power and human agency.
Filtered records
AI & Humanity · 6 min read
In narrow, well-measured domains AI already outperforms people; the open question is whether accuracy on a defined target is the same thing as good judgement.
ReadAI & Humanity · 6 min read
Automation without human approval is defensible where the decision is reversible, the stakes are bounded and a real appeal route exists — and hard to defend where any of those is missing.
ReadAI & Humanity · 6 min read
Over-trust turns a system's errors into social facts: once an output is treated as the default account of reality, correcting it requires more evidence than producing it did.
ReadAI & Humanity · 6 min read
It does not need to fabricate anything. Ranking, summarising and repetition are sufficient to make some accounts feel obvious and others feel marginal.
ReadAI & Humanity · 6 min read
Malice is containable because it can be identified and opposed; indifferent competence is not, because there is nothing in it to appeal to.
ReadAI & Humanity · 6 min read
Confidence is persuasive independently of accuracy, so a system that presents unhedged answers will out-argue a careful human expert who says the situation is complicated.
ReadAI & Humanity · 6 min read
Automation bias is the tendency to accept a system's output rather than scrutinise it — and it is the main reason human-in-the-loop safeguards weaken over time.
ReadAI & Humanity · 6 min read
Paper is a worse record in nearly every respect except the one that counts in a dispute: it does not silently change, and a change to it leaves a mark.
ReadReader record
Get release news, cover reveals, exclusive extracts, author notes and first access to the next story.