When we search for the humanity in LaMDA AI, we’re missing the point
In the Globe and Mail on June 25, 2022, some commentary on the Blake Lemoine/sentient AI episode:
[snip]
The whole exchange is essentially a version of the famous “imitation game” proposed in the 1950s by mathematician Alan Turing and designed to see whether a machine, writing responses to human questions, could “pass” as a human. But there has been a bit of a misunderstanding along the way around Turing’s intent: The test was not designed to show whether machines were capable of human-like thought.
In fact, Turing considered the question “can machines think?” to be “too meaningless to deserve discussion.” For him, a far more interesting question was whether machines could use language to trick an interrogator into thinking it was human. The Turing test, then, is intended to be an inquiry into human suggestibility, rather than some barometer of machine intelligence.
[snip]
Click here for the full article.