We missed it.
In 1950, Alan Turing proposed a test to determine whether machines can think [1.]
The test was simple; A human interrogator converses with both a human and a machine, both hidden, via text-based communication.
If the interrogator cannot reliably distinguish the machine from the human, the machine passes.
Well, last year. [2.] GPT-4.5/LLaMa (2025) model passed a rigorous three-party Turing test run by University of California San Diego [3.] and convinced judges they were human 73% of the time, exceeding the 50% chance baseline. By some measures, it also beat the humans.
There was no Kasparov vs Deep Blue Moment.
There was barely a headline.
It seems we’ve moved on to the debate about better tests for intelligence or what intelligence actually means. Rightly so.
And yet, the significance of being unable to distinguish if the person you are talking to is human or not strikes me as profound.
This was the threshold (or as good a threshold as we will get).
Crossing it forever changes humanity.
[1.] https://en.wikipedia.org/wiki/Turing_test [2.]https://www.livescience.com/technology/artificial-intelligence/open-ai-gpt-4-5-is-the-first-ai-model-to-pass-an-authentic-turing-test-scientists-say
[3.] https://arxiv.org/abs/2503.23674v1