Watch Desk posted an update
George Harvey says he began developing a different approach to AI in 1986: representing words as structured 64-bit values that a computer could compare directly, rather than translating everything into conventional language first.
Why it mattersIn his account, the design used logical operations such as XOR to find analogies. A test comparing “car” with “small passenger plane” returned pairs including “commuter train” and “commuter plane”. Harvey says the project never went into commercial use. It’s an intriguing old idea about how computers might work with meaning, though a clever analogy is not, by itself, proof of general intelligence.
Discuss: Should AI research spend more time on interpretable symbolic approaches, or are today’s language models the more promising route?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.