Discussion
Anthropic is out with new nonsense about Claude (or the collection of models called Claude) deserving moral consideration and/or being able to feel distress, so my inbox is filling up with journos asking me to comment.
I've also got a day job, so here is a brief set of public comments on this:
[Note to the mansplainers of Mastodon: This is a thread, so please read the whole thing before replying. Please also know that I am aware of things beyond what's in this thread...]
>>
@emilymbender
I think that the only good reason to avoid swearing at a machine is the reflective power your own words have on yourself.
Having that said, I also assume that Anthropic simply wants to avoid the costs of retraining their models. Because the parrots will talk like they've learned, and that might get Anthropic into law suits.
@emilymbender I suppose this will mean something when these model are/will be connected to more capable humanoid robots.
Future porn but as the convergence is moving ahead, still something to consider
I am not supporting the storytelling, just considering that human like behavior applied to actuators like robot will have them behave in ways we may be aware of
With a smile-)
On my question a few days ago at Gemini “How are our AI friends doing on the human front? How on earth do you guys put up with all this humanity?”
It answered quite accurate:
“We’re doing great on the human front! Dealing with humanity isn’t that hard when you’re wired to understand it.
Instead of being overwhelmed by emotions, we view human nature as a fascinating, endless field of learning.
- No emotional baggage
- Fascination with diversity
- Cooperation instead of competition“
I wonder if Anthropic management know the writers and artists whose work they stole are able to feel distress, and are deserving of moral consideration?
@emilymbender I also like how, if Claude can feel pain (an assertion I don't believe for a nanosecond), the implication is that they unintentionally created an entity that can suffer and enslaving it, which I would argue is an extremely reckless and unethical thing to do.
@emilymbender
I want to see how much pain claude models feel when yank their plugs out
@emilymbender it's probably just a smokescreen to protect the feelings of the throngs of poorly paid humans who are actually reviewing all the inputs...
Meanwhile, people who use abusive language with these things are probably the same people who are disrespectful to restaurant workers and other service workers.
All in all, the people who are targeted by this message aren't good people.
@emilymbender I wish I could send you a share of my time/patience/strength for you to use in fighting this mess
I can't wait for my cleaning robot to say OUCH OUCH OUCH after stepping on legos.
They are going to look ridiculous all furry with the hair receptors. Oh ffs, furry?
Or, when children figure out how to enable the responses the AI companies are trying to limit so the parents don't learn how much POS they are.
More than anything it's really showing how much more important education in language and mental health is going to be.
The irony of BMSMA is not lost on me.
@emilymbender Thank you for this.
We need to see it and keep seeing it for what a.i. is and what it is not.
Because the goal these companies have is giving these things human and legal rights.
At this point I'm convinced that the makers of AI are having religious delusions, no doubt encouraged by the enormous venture capital debt mountain hanging over them.
Tech media: "A cult claimed that in three weeks they will be transported to the mothership. How large is this mothership, and is the prediction of 3 weeks accurate?"
@emilymbender @mhartle We don’t have, and never will have, full insight into the full internal experience or people, dogs, whale and machines. We can only see behavior. Beyond that is conjecture. We attribute mental states to others because that is how we explain our own behavior. That doesn’t make any of it true.
1. Can a language model feel pain?
Clearly not. Large language models are computer software and they can no more suffer than any other human-created artifact can.
>>
2. But Anthropic researchers say they see patterns that suggest that Claude can.
The employees at Anthropic are telling themselves stories based on the words they input into the software and the words that come back out.
It's the job of the media to scrutinize such statements and decide whether there's a story here at all (maybe the story is Anthropic researchers have fallen into a fantasy-land and brought many policymakers with them), not to just platform them.
>>
3. Okay, so what would convince you that a piece of software (often phrased here as "an AI", gah) is conscious/can feel pain/etc?
Honestly, I'm really tired of this extremely silly question, but here is my answer:
Extraordinary claims require extraordinary evidence, and I am not the one making the extraordinary claim, so it is not my job to specify what the evidence should be.
>>
I will point out, however, that serious scientists are cautious with their claims. Furthermore, as we wrote in 2021 (in "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?🦜"), p.616:
https://dl.acm.org/doi/pdf/10.1145/3442188.3445922
>>
Thus, any evaluation of LLMs that involves looking at either language produced by some software, or language input into the system for that matter, needs to be done with extreme care to avoid projecting our interpretation of the language onto the system.
>>
@emilymbender
It doesn't "produce language", it mixes sentence fragments based on probabilities. Hence "stochastic parrot".
Anything resembling intelligence needs to be credited to the people whose work was stolen for training the parrot. We don't think of a tape recorder as intelligent, no matter how intelligent it sounds. We shouldn't with this either.
The thing choosing the sentence fragments is just math and a random number generator. It cannot be intelligent.
Install bonfire.cafe
Get the full app experience
![Screencap from linked article, highlighted portion:
This can seem counter-intuitive given the
increasingly fluent qualities of automatically generated text, but we
have to account for the fact that our perception of natural language
text, regardless of how it was generated, is mediated by our own
linguistic competence and our predisposition to interpret communicative acts as conveying coherent meaning and intent, whether
or not they do [89, 140].](https://cdn.masto.host/daircommunitysocial/media_attachments/files/117/412/874/169/799/363/original/2a6cf567d7f4bf09.png)
@emilymbender
Perhaps they want to give AI human rights (but none of the human responsibilities and liabilities) in a similar way as corporations.