BS Summary: This article contains 7 faulty reasoning types, including Appeal to Emotion, Framing Effect, and Appeal to Authority, with Hasty Generalization as the most egregious example at 30.8% saturation with 100 hits. Analysis detected 345 faulty-reasoning hits from 325 analyzed words, generating a BS Score of 47.5% and a BS Rank of 45% (11,807 of 21,185 articles). This article is better (less manipulative) than 55.70% of the article peer group.
Talking about artificial intelligence has become its own industry.
AI will save us or doom us.
It will cure cancer or end work.
Panels, manifestos, congressional hearings, op-eds without end — and through all of it, the machines keep getting better at talking back, and no one can tell you whether the thing you confided in last night was good for you or not.
Our guest on this week’s WhoWhatWhy podcast, Erika Anderson, has stopped arguing and started measuring.
She is an unlikely person to be grading the machines.
Not an engineer.
A writer — born on a commune, trained in the personal essay, drawn since childhood to the question philosophers have been chewing on for 3,000 years: What does it mean to be human?
That question turns out to be the whole ballgame now, and she may be better equipped for it than the people writing the code.
What she built is called HumaneBench.
It doesn’t measure whether a chatbot is smart.
It measures how the chatbot treats you — your attention, your dignity, your relationships with actual human beings.
Eight principles, applied to the thing millions of us now talk to at two in the morning.
Is it early?
Yes.
Is it crude?
Anderson would say so herself — she is disarmingly candid in this conversation about where the whole enterprise could break, invoking everything from Sherry Turkle to Jurassic Park to Heisenberg to explain why measuring a living system is harder than it looks.
She has even turned the instrument on her own company and published what came back.
In this conversation Anderson takes on the loneliness economy, the business model that profits from your dependency, what happened when OpenAI retired GPT-4 and users reacted as though someone had died, and the question nobody in Silicon Valley wants pinned down: Where does personal responsibility end and corporate accountability begin?
It’s a start.
Someone had to go first.
Analysis
Hover over highlighted words in the article to view the associated bias or fallacy analysis.