
<aside>
💬
"The history of AI is full stop a history of a lack of faith in people."
</aside>
Thursday October 29, 2026 · 4:05 PM · Talk
Key ideas from the talk
- AI began after the Second World War, when military planners doubted that people could be trusted to run Cold War decisions.
- First paradigm: symbolic AI at the RAND Corporation. It assumed intelligence is reasoning by rules. It failed outside chess and logic.
- Second paradigm: expert systems. Knowledge engineers turned experts' knowledge into if-then rules. TurboTax is her example.
- Critics argued that human intelligence grows from emotion, embodiment and the mix of rational and irrational.
- The third turn skipped modelling ourselves. Language models grew out of natural language processing, which moved from grammar rules to word statistics in the 1980s.
- She says we still oversell narrow definitions of intelligence. She invites people to fight for forms of intelligence that celebrate human experience.
- In the Q&A she said science is a messy human endeavour, and that consciousness is still not understood.
Quotes
- "All the data that trains our artificial intelligence systems is data from the past."
- "We're too slow. We're too hungry. We're too superstitious. We're too arrogant. We're too emotional. We are too tired to be trusted."
- "What if human intelligence develops through and because of all of these facets of the human experience that became so unworthy of our trust after the Second World War?"
- "Could we bypass ourselves altogether?"
- "Part of what we're fighting for right now is the very definition of intelligence itself."
- "Trusting each other is a political act for so many reasons."
- "Scientific practice is very messy. It is a human undertaking."
About the speaker