2 ms·
The Chinese Room thought experiment is kind of a different mistake. In the Chinese room argument, Searle attempted to imagine an AI system that was "obviously n
by hnfong 4y ago
The Chinese Room thought experiment is kind of a different mistake. In the Chinese room argument, Searle attempted to imagine an AI system that was "obviously not sentient" because it was made up of hard rules and facts (presumably legible to us humans), and it was argued that it should be obvious that the system wasn't sentient.
For the ChatGPT case it's not clear the Chinese room argument even applies. First, it's not even a Chinese room, it's trained by a neural network, an a neural network is inspired by the structure of our brains. There's no "set of instructions in English" for anyone to follow, but it's a rather overwhelming set of neurons somehow producing the results that it was trained on, and as an outsider we have no idea what's going on inside individual neurons -- just like our brains.
I'm seeing a lot of people claim that "we know how ChatGPT is created, and so it's not sentient", maybe this looks like the Chinese room fallacy (I'm calling it as such), but it's like claiming we know the laws of physics (Schrödinger's equation etc.) and then claiming we know how the brain works because it's all quantum physics. The fact is we know how to build GPT but we don't actually know how it works.
Besides what happens to humanity when we learn more and more about how our own brains work? I'm pretty comfortable with the idea that we're not sentient (yay), but I'm pretty sure those same people would then come up with other nonsensical arguments to prove that they are different from the robots we created.
- larryfreeman 4y agoI believe that it is straight forward to show how Searle's Chinese Room argument can be applied to ChatGPT. The idea is not the "process" for building the rules but the "process" where the rules are applied. True sentience requires a "true" understanding of the information that is being presented (otherwise, it is not reflective of sentience but rather trickery). The Chinese Room applies because ChatGPT or any other computer is generating text purely through logic, that is, logical rules. To make this point clear, let's assume just for argument's sake that at some quantum level, the electrical activity underlying a computer is partially sentient (I am not making this claim -- just using it as an example). Even if this were true, this would not prove that the sentience is involved with the text being generated by the same computer. This underlying sentience would be like the "person" in the Chinese room. The text generated are the Chinese words which would be completely independent from the "sentience" following the rules. Generating text at sufficient complexity to fool the average person does not prove that "sentience" is occurring. Generating text consistently in a way that is only explainable through sentience is what is required. ChatGPT, because it is easily shown to be flawed, does not rise to this level of evidence. In the Chinese Room example, the complexity of what could be displayed by the rulebooks would have a limited complexity (though, the surprise of ChatGPT is that this complexity is greater than many of us would have supposed). Just like the story of Clever Hans the Horse (see https://en.wikipedia.org/wiki/Clever_Hans https://en.wikipedia.org/wiki/Clever_Hans), once the limitation of the process is realized, it becomes clear that there is not "true" understanding of what is generated which means that the rules, in themselves, do not reflect "true" sentience.