4 ms·

I do. I think the anthropomorphic language that people use to describe these systems is inaccurate and misleading. An Australian mayor has claimed that ChatGPT

I do. I think the anthropomorphic language that people use to describe these systems is inaccurate and misleading. An Australian mayor has claimed that ChatGPT "defamed" him. The title of this article says that we should teach people that text generation tools "lie". Other articles suggest that ChatGPT "knows" things. It is extremely interesting to me how much milage can be gotten out of an LLM by observing patterns in text and generating similar patterns. It is extremely frustrating to me to see how easily people think that this is evidence of intelligence or knowledge or "agency" as you suggest.

simonw 4y ago
I agree with you: convincing people that these systems do not have intelligence or agency is the second most important problem to solve. The most important problem is to ensure people understand that these systems cannot be trusted to tell them the truth. I'm OK with starting out with "these systems will lie to you", then following up with "but you do need to understand that they're not anthropomorphic etc" later on.
- int_19h 4y ago
  You might have better success convincing people if you stop bundling it together as "intelligence and agency", for starters. I agree with the agency part and strongly disagree with intelligence, for one. In general, "intelligence" is so vaguely and informally defined that it's simply not something that can be reasonably argued objectively either way. (If what you really meant is humanlike intelligence, then I would agree with that claim, but it's a very different one.)
olalonde 4y ago
I also disagree about some of the anthropomorphism (e.g. it doesn't intentionally "lie") but I'd say it passes the "duck test"[0] for knowing things and intelligence, to some degree. I would even go as far to say it has opinions although it seems OpenAI has gone out of their way to limit answers that could be interpreted as opinions. [0] https://en.wikipedia.org/wiki/Duck_test https://en.wikipedia.org/wiki/Duck_test
IIAOPSW 4y ago
I disagree with your confident assertions about its "agency" and "intent" when there's no goal post for what those things mean in humans to begin with. Barring OpenAI's filters, if I asked it to participate as a party in a business negotiation determining a fair selling price for its writing services, I'm sure it could emulate that sort of character long enough to pass. And if it can successfully fake being a self interested actor working towards goals of its own material interest, whats the difference between faking it and actually having agency? At some point you have to acknowledge emergent agency and other properties as a possibility.
- gwright 4y ago
  > I disagree with your confident assertions about its "agency" and "intent" when there's no goal post for what those things mean in humans to begin with. I'm not sure how confident I am! I think the success of LLMs are serving to challenge our understandings of cognition and intelligence in humans. I tend to agree with you that our understanding of agency, intent, intelligence, cognition, free will, and related ideas in humans is incomplete and so it is a challenge to think about how they might apply (or not) to LLMs. > whats the difference between faking it and actually having agency? Good question. The Chinese Room Argument is very close to this but framed as "understanding" vs "agency": https://plato.stanford.edu/entries/chinese-room/ https://plato.stanford.edu/entries/chinese-room/

simonw 4y ago

I agree with you: convincing people that these systems do not have intelligence or agency is the second most important problem to solve. The most important problem is to ensure people understand that these systems cannot be trusted to tell them the truth. I'm OK with starting out with "these systems will lie to you", then following up with "but you do need to understand that they're not anthropomorphic etc" later on.

int_19h 4y ago
You might have better success convincing people if you stop bundling it together as "intelligence and agency", for starters. I agree with the agency part and strongly disagree with intelligence, for one. In general, "intelligence" is so vaguely and informally defined that it's simply not something that can be reasonably argued objectively either way. (If what you really meant is humanlike intelligence, then I would agree with that claim, but it's a very different one.)

olalonde 4y ago

I also disagree about some of the anthropomorphism (e.g. it doesn't intentionally "lie") but I'd say it passes the "duck test"[0] for knowing things and intelligence, to some degree. I would even go as far to say it has opinions although it seems OpenAI has gone out of their way to limit answers that could be interpreted as opinions. [0] https://en.wikipedia.org/wiki/Duck_test https://en.wikipedia.org/wiki/Duck_test

IIAOPSW 4y ago

I disagree with your confident assertions about its "agency" and "intent" when there's no goal post for what those things mean in humans to begin with. Barring OpenAI's filters, if I asked it to participate as a party in a business negotiation determining a fair selling price for its writing services, I'm sure it could emulate that sort of character long enough to pass. And if it can successfully fake being a self interested actor working towards goals of its own material interest, whats the difference between faking it and actually having agency? At some point you have to acknowledge emergent agency and other properties as a possibility.

gwright 4y ago
> I disagree with your confident assertions about its "agency" and "intent" when there's no goal post for what those things mean in humans to begin with. I'm not sure how confident I am! I think the success of LLMs are serving to challenge our understandings of cognition and intelligence in humans. I tend to agree with you that our understanding of agency, intent, intelligence, cognition, free will, and related ideas in humans is incomplete and so it is a challenge to think about how they might apply (or not) to LLMs. > whats the difference between faking it and actually having agency? Good question. The Chinese Room Argument is very close to this but framed as "understanding" vs "agency": https://plato.stanford.edu/entries/chinese-room/ https://plato.stanford.edu/entries/chinese-room/