5 ms·
This is relevant, I'm surprised it's downvoted. Sure, it's a bit scoldy, but it's not wrong. Downvote irrelevant content, not content you disagree with. The
by DanHulton 4y ago
This is relevant, I'm surprised it's downvoted. Sure, it's a bit scoldy, but it's not wrong.
Downvote irrelevant content, not content you disagree with.
The tendency to personify these chatbots when they're just statistical text generators and _not_ people is interesting but also potentially troubling. On the one hand, we do this all the time to really anything with a face (people of all ages will anthropomorphize stuffed toys, for example) so it's not _surprising_ that something that seems to exhibit a personality gets the same treatment.
But on the other hand, we tend to anthropomorphize stuffed toys in a playful way because we know they can't possibly be alive. There seems to be a broader lack of clarity about what these chatbots actually represent -- are they just one or two steps away from true sentience? They're absolutely not, but in the broader conversation, questions like this are being thrown around. And then to come on here and see casual anthropomorphization, it kind of makes you wonder how far these attitudes are actually reaching, despite HN being a more-technical audience that should have the capacity to understand that there's no THERE there, it's all just clever statistics.
- Mistletoe 4y agoHow do you know your brain isn’t just clever statistics and a language model where it picks the next word? I talk to lots of normal people and they aren’t really that different than this. Regurgitating something they saw on the news coupled with whatever logical fallacies and biases they have accumulated in their lifetime and from their environment that they want to throw in. Yet we are so certain they are sentient…
- goatlover 4y agoBecause LLMs are something humans created to generate text based on existing human language. You’re using a metaphor and mistaking it for the real thing. People have done this before with technology, because we don’t have a good understanding of how the brain works.
- xyzelement 4y agoI don't think the person you are responding to is stretching a metaphor. He's pointing out human behavior that might just as well have been produced by an LLM. I think that's a correct observation.
- goatlover 4y agoHumans have bodies that interact with the world. We have to concern ourselves with things like survival, relationships, and what not. An LLM is just a tool for generating human-like text. It's not biological, and doesn't have any drives. It's not trying to hookup, obtain food, or climb the social ladder.
- xyzelement 4y agoI think you are just saying 'humans aren't machines running LLM' and nobody is arguing with you. There's a subtler thing being pointed to here.
- goatlover 4y agoI'm saying humans don't produce language the same way LLMs do, because we have biological and social factors that result in the langue we use. Which is a bit more than predicting the next word based on a statistical model, even if that model does a good job at generating text that looks human-like. So yes, LLMs incorporate a lot of the structure of human language. No, that's not quite the same thing as humans producing language.
- layer8 4y ago> Downvote irrelevant content, not content you disagree with. https://news.ycombinator.com/item?id=117171 https://news.ycombinator.com/item?id=117171
- luckylion 4y agoThey are presenting as persons, usually including a name and gender. It's hard to argue that they don't express gender identity. Using the correct pronouns is suddenly taboo? It's strange to me that people are demanding to call them 'it'.
- pmlnr 4y agoThey do not have gender, nor identity. They are not people.
- luckylion 4y agoCan you identify a specific bot when you talk to them repeatedly? Isn't that identity? Do they present a gendered name? Then why not use it to refer to them? What's bothering you about it?
- pmlnr 4y agohttps://news.ycombinator.com/item?id=35024508 https://news.ycombinator.com/item?id=35024508
- luckylion 4y agoWhat's the problem with people getting attached to them? People get attached to all kinds of things including the pen their late mother gave them, or that pair of pants they wore when they were younger. Lovecraft's gods were operating outside the laws of our universe, that's what made them terrifying. I don't think chatbots will be able to do that.