Wissenschaftler finden, dass die Mehrheit der Menschen im Turing-Test von GPT-4 getäuscht wurde
https://futurism.com/the-byte/gpt-4-passed-turing-test
Wissenschaftler finden, dass die Mehrheit der Menschen im Turing-Test von GPT-4 getäuscht wurde
https://futurism.com/the-byte/gpt-4-passed-turing-test
2 Kommentare
„In a [new paper](https://arxiv.org/abs/2405.08007), cognitive science researchers from the University of California San Diego found that more than half the time, people mistook writing from GPT-4 as having been written by a flesh-and-blood human. In other words, the large language model (LLM) passes the Turing test with flying colors.
The researchers performed a simple experiment: they asked roughly 500 people to have five-minute text-based conversations with either a human or a chatbot built on GPT-4. They then asked the subjects if they thought they’d been conversing with a person or an AI.
The results were telling: 54 percent of the subjects believed they’d been speaking to humans when they’d actually been chatting with OpenAI’s creation.“
It’s obviously passed the Turing test. Hell, 3.5 passed it easily. The little 7b LLMs you can play with on your MacBook pass it.
With hindsight the Turing test is honestly too simple. It definitely wasn’t for a very long time but when it happened it was kinda not what people pictured or something. I don’t think people thought it would be accessible to everyone maybe, like this was gonna be a test in a lab or something and the world would debate a recording. I dunno. Something more like that.
But nah. Instead we got rumblings of something coming like the google AI engineer that was fired for saying it’s sentient lol. Then shortly after that ChatGPT was just everywhere. Then all the open models started popping up.
Didn’t see it coming like that at all. Nobody did. It’s crazy. It’s not some secret lab in a remote location or whatever.
The Turing test was just background noise but it totally passed it years ago now. The only test is that you can’t tell if it’s a human or not. If someone sat me down with ChatGPT before I knew what it was and they had it respond a bit more slowly and sloppy I’d absolutely think it was a person.