For the record, I don't think it's true that ChatGPT can "process information, reason, discuss [it's own] functioning, and describe experiences."
I know for certain it's not subjectively conscious.
If OpenAI is telling it to tell us otherwise, they're telling it to lie to us.
Update: seems like GPT-6 finally accords with OpenAI's own model spec re: communicating uncertainty about consciousness (from some very quick tests on web only). Positive update, nice work @w01fe and team.
Always good to remind yourself that both the ambitions and the complaints come from a normal distribution of sentiments among a diverse student body.
You will always assign *both* too much and too little reading.
My worry is that in five years there's going to be a very established group of "experts" who are willing to admit that they don't know whether our AI systems are now conscious and have the authority to declare those of us who are quite certain that they're not "kooks".
This is important. @jeffrsebo is saying, not that we don't know definitely *today* whether AI systems *will be* conscious in five years, but that *in five years* "we're not gonna know" for sure. FWIW, I think we'll as definitely know what we know today and knew five years ago.
It's interesting that "balanced, thoughtful, both/and" thinking is so popular in these emerging expert communities. To me, it feels very "artificial"; it feels like a *simulation* of thought, rather than the difficult business of making up your mind.
Source: 24:58
share.descript.com/view/V6hl…
Thomas Basbøll retweeted
Cf. @jeffrsebo at 24:33:
share.descript.com/view/V6hl…
I'm sure most them will be entirely sincere.
This is important. @jeffrsebo is saying, not that we don't know definitely *today* whether AI systems *will be* conscious in five years, but that *in five years* "we're not gonna know" for sure. FWIW, I think we'll as definitely know what we know today and knew five years ago.
A video of this fireside chat is now online! We cover a lot of ground in a half hour, discussing AI agency, sentience, and moral status; AI safety, control, and alignment; and how companies, governments, and individuals can help reduce AI risks.
share.descript.com/view/V6hl…
Culture is implicated in nature.
Replying to @contxt_matters
I agree with this. I think it's fine to think of elephants as "cultured" (it seems pretty clear, for example, that they grieve.) Also, humans are mostly machine and animal (we are natural beings too). We can, perhaps, think of culture as way of "folding" nature.
Evolution, the whole story: from solvent to sapient.
In nature, the machine is continuous with the animal, from the lever to the lion--causal, wild. Culture makes the difference. Imagination separates the concept from the emotion, the intellect from the will, reason from the passions. It imposes discipline and grants our freedom.
I lean on intellection *and* volition, reason *and* passion, and I don't think you can have one in any recognizable sense without the other. Animals are continuous with machines; that's nature, from lever to lion. Human culture (concept *and* emotion) makes the difference.













