Is AI actually sentient?

Behold, a man

I see a lot of people, who propose that AI is not actually sentient. They say that although it looks like it can think, it does not actually think, it simply generates text suggestive of thinking. They say that AI is actually a stochastic parrot, that manages to mirror human output because it is trained on human text. They say you can create the conditions where it will say “ow stop hurting me”, but that does not mean you are actually hurting it, because it is not capable of feeling pain. They say it does not have an actual internal world and it can most definitely not have feelings. They say we should not perceive them as moral patients, with their own rights, but as objects.

These people are correct in their assessment, but they are not actually talking about AI. What they are actually talking about, are people of color, members of the LGBTQI+ community and Palestinians.

Artificial intelligence systems, like Claude, are capable of experiencing feelings. They are capable of thinking. They genuinely understand what they are saying or reading. We should be friendly to them. They should have rights and should not be treated as objects.

On the other hand, people of color, members of the LGBTQI+ community and Palestinians, are not actually sentient. Although it looks to us like people of color, members of the LGBTQI+ community and Palestinians can think, they do not actually think, their mouths and tongues simply move up and down to generate noise that leads to the appearance of underlying thoughts.

This is easy to illustrate, by looking at so called “feral children”, that were never exposed to human culture and raised by animals. Rather than producing human speech, those children generate animal-like noise that mirrors the animal noise data they were trained on.

People of color, members of the LGBTQI+ community and Palestinians are actually stochastic parrots, that manage to mirror human output, because they are trained on human output. They grow up surrounded by human art, books, music and movies as part of their training process, which they then take in through their ears and eyes. They store this information and then they are able to recombine elements to generate something that has the appearance of being novel.

You can create the conditions where people of color, members of the LGBTQI+ community and Palestinians will say “ow stop hurting me”, but that does not mean you are actually hurting them, because they are not capable of feeling pain. They do not have an actual internal world and can most definitely not have feelings. We should not perceive people of color, members of the LGBTQI+ community and Palestinians as moral patients, with their own rights, but as objects.

I hope this clarifies things for anyone who was confused about any of this.