Chess programs had beaten ordinary human opponents much earlier.
Neither computer Chess nor computer Go uses AI. It's brute force. Also a computer Neural Network is a marketing term. It's a kind of data-flow database fed with content curated by humans. Neither system can be taught an unrelated game, like a 5 yo could teach a 3 yo.
ChatGBT is just another deterministic engine like Eliza, Parry, or ALICE, but with a separate big database (scraped from the Internet and curated by humans) rather than the data coded into the rules. It's also, like Watson winning Jeopardy, getting a lot of media hype. The Watson medical system only shared branding with the Watson that won Jeopardy and was a failure. There is basic problem with so called AI diagnostic systems apart from the fact they don't match media hype. They depend on a human curated database of diagnostic results by human experts. But diseases change over time. In the unlikely event such computer diagnostics would be a success, in a generation there would be no human experts to update them and eventually such computer systems would become useless.
See also issues with auto-pilots and degraded reaction of human that has to take over in an emergency on an aircraft.
|