Post · bonfire.cafe

@emilymbender@dair-community.social · yesterday

Usually, when I get interviewed for a piece on something like "AI consciousness" I am relegated to the skeptics box --- some short paragraph near the end. So it is a nice change to see this piece by Holly Baxter

https://www.the-independent.com/tech/ai-news-humanize-chatbot-conscious-b2963788.html

🧵>>

The Independent

The people building AI think it might be conscious. That’s not the most alarming part

Anthropic’s CEO Dario Amadei says he can’t rule out that its chatbot, Claude, is conscious. A Google engineer is sure he once built a sentient being. Holly Baxter speaks to the experts about whether or not ‘AI welfare’ is a serious pursuit — and what that means for humans

Gemma ⭐️🔰🇺🇸 🇵🇭 🎐

@gcvsa@mstdn.plus · 14 hours ago

@emilymbender If you ever wanted to know how religion got started in human civilization, here it is, playing out in real time. Make it spooky, make it hype.

J. R. DePriest :verified_trans: :donor: :Moopsy: :EA DATA. SF:

@jrdepriest@infosec.exchange · yesterday

@emilymbender

When I read that headline, it gave me the impression that "AI" was going to be declared as more than conscious in some way. I suppose that's just "how you write a headline".

I was pleasantly surprised at how sober Holly Baxter's take on "AI" was. She does not blindly buy in to the hype and she hasn't fallen down the rabbit hole of installing Claude and getting bamboozled by its magical cold reading skills.

I was further surprised to see just how much space was given over to your interview.

Thank you for even taking the time to continue talking to reporters when, as you said, you are often a checkbox just so they can say they did a "both sides".

Thalia Archibald

@thalia@discuss.systems · yesterday

@emilymbender You mention a $2,000/hr consulting fee. Are you also getting a flood of prospective students you have to turn away?

George B

@gbargoud@masto.nyc · yesterday

@emilymbender

> a message specifically included for tech bros with startups who want to download all her knowledge about LLMs: “My consulting fee is $2,000/hour. I do not ‘grab coffee’ or ‘jump on the phone’.”

Nice, how many of them took you up on that?

cstrauber

@cstrauber@mastodon.social · yesterday

@emilymbender It is *fascinating* how you appear in AI-related media. Smart reporters and tech people know they have to mention you, but they can't engage with your arguments without turning off the hype machine. Thanks for sharing this.

spdrnl

@spdrnl@sigmoid.social · yesterday

@emilymbender The real issue here might be that machine learning models pursue a single objective; no matter what.

So the next step is then to say that machine learning models are superior because of the single objective.

Being human means not having a single objective. These people are rich and powerful enough to redeclare utilitarianism

It's all a bit narrow minded; one long impaired intellectual gooning session.

Prof. Emily M. Bender(she/her)

@emilymbender@dair-community.social · yesterday

@spdrnl Good ol' Mastodon. First reply is of course some more mansplaining.

Save Elderly Housing! Patty 👀

@pattykimura@beige.party · yesterday

@emilymbender

❤

@spdrnl

Don't ever mansplain to a an internationally known subject material expert whose consulting fee schedule for tech bros starts at $2000 per hour.

spdrnl

@spdrnl@sigmoid.social · yesterday

@emilymbender Oh, that was not my intention.

I was reacting to the caption under the photo; I highly distrust the AI crowd. My post was meant as an inside take, what I think is behind these projections.

My statement was intended to sympathize with your many good ideed; I really admire your take on things.

I can remove the post.

Prof. Emily M. Bender(she/her)

@emilymbender@dair-community.social · yesterday

@spdrnl My advice is if you want to do something like that, make it clear in your post who you are addressing your comments to.

You started by clicking "reply" to me, so the default interpretation is that you're replying to me.

Another option is to quote post instead. Or post your own link to the article.

1+ more replies

Prof. Emily M. Bender(she/her)

@emilymbender@dair-community.social · yesterday

I have been sharing the Magic 8 Ball analogy for a while now, but I think this is maybe the first time it's made it to print:

2 media

Screencap from linked article:

Our human tendency to apply context to language and imagine that it’s coming from a personality primes us to believe in AI consciousness, says Bender. And when we interact with language, we start shaping our inputs to match the linguistic outputs: we convince ourselves further by adapting. She now keeps a Magic Eight ball in her office at all times to demonstrate this exact point: “When you play with the Magic Eight Ball and you ask it ‘Should I bring an umbrella?’ and it says something like ‘Signs unclear, ask again later,’ then OK. That would work for any question. But if it says ‘Without a doubt’, which is what came out right now, that only works if it was a yes/no question. If I say ‘What should I have for lunch?’ and I get back ‘Without a doubt’, that's incoherent. And so in playing with this toy, we sort of learn to shape the inputs that we give it so that we can make sense of the outputs that come back.” — Screencap from linked article: Our human tendency to apply context to language and imagine that it’s coming from a personality primes us to believe in AI consciousness, says Bender. And when we interact with language, we start shaping our inputs to match the linguistic outputs: we convince ourselves further by adapting. She now keeps a Magic Eight ball in her office at all times to demonstrate this exact point: “When you play with the Magic Eight Ball and you ask it ‘Should I bring an umbrella?’ and it says something like ‘Signs unclear, ask again later,’ then OK. That would work for any question. But if it says ‘Without a doubt’, which is what came out right now, that only works if it was a yes/no question. If I say ‘What should I have for lunch?’ and I get back ‘Without a doubt’, that's incoherent. And so in playing with this toy, we sort of learn to shape the inputs that we give it so that we can make sense of the outputs that come back.”

Screencap from linked article:

When people play around with Claude or some other chatbot, they’re doing the same thing, Bender adds: “We are putting in input that allows us to contextualize and shape how we understand the output that comes from.”

As for consciousness, she says, to continue her analogy: “Imagine you took a Magic Eight ball and instead of a 16-sided die or whatever inside of it, you made it really big. So you could have a 256-sided die. And then you filled up a football field with those. Is that any closer to consciousness than the one little one that I have in my hand here?” — Screencap from linked article: When people play around with Claude or some other chatbot, they’re doing the same thing, Bender adds: “We are putting in input that allows us to contextualize and shape how we understand the output that comes from.” As for consciousness, she says, to continue her analogy: “Imagine you took a Magic Eight ball and instead of a 16-sided die or whatever inside of it, you made it really big. So you could have a 256-sided die. And then you filled up a football field with those. Is that any closer to consciousness than the one little one that I have in my hand here?”

Robin Adams

@robinadams@mathstodon.xyz · 22 hours ago

@emilymbender This company is selling a magic 8-ball as "Offline ChatGPT":

https://www.spellingmistakescostlives.com/single-post/chatgpt-magic-8-ball

SPELLING MISTAKES COST LIVES

CHATGPT MAGIC-8 BALL

After much research and development I have finally made an offline version of ChatGPT. Now you can save water and electricity while carrying one of the world's most powerfully annoying AI chatbots in your pocket. Have every whim affirmed with up to 20 of the most popular ChatGPT responses. Smooth your brain into a frictionless hypermind capable of instant regurgitation via a corporate flattery and theft engine. They're 40 quid, and you can order one here. I have a limited pre-Xmas supply with mo

Hanno Zulla

@hzulla@infosec.exchange · yesterday

@emilymbender When I explain my qualms about GenAI chatbots to others, I usually refer to Clever Hans as a historic example of a situation where an observer falsely attributes "intelligence" to a non-intelligent process.

https://en.wikipedia.org/wiki/Clever_Hans

Clever Hans - Wikipedia

Christian Pietsch

@chpietsch@fedifreu.de · yesterday

@emilymbender I did not know what a Magic 8 Ball is, so I looked it up: https://en.wikipedia.org/wiki/Magic_8_Ball

Magic 8 Ball - Wikipedia

Prof. Emily M. Bender(she/her)

@emilymbender@dair-community.social · yesterday

"Technologies of isolation" is due to @hypervisible , but even if I'm careful to tell journalists that, they don't necessarily include the attribution.
>>

Screencap from linked article:

There’s a reason why these are referred to as “technologies of isolation,” she adds: because it simply isn’t a neutral choice to talk to Claude or Grok or ChatGPT about your loneliness. It is a choice that ends up “weakening your connections” and driving you further away from the human beings who could actually help you. — Screencap from linked article: There’s a reason why these are referred to as “technologies of isolation,” she adds: because it simply isn’t a neutral choice to talk to Claude or Grok or ChatGPT about your loneliness. It is a choice that ends up “weakening your connections” and driving you further away from the human beings who could actually help you.

Prof. Emily M. Bender(she/her)

@emilymbender@dair-community.social · yesterday

I also appreciated O'Niell's comments quoted in the article:

Screencap from linked article:

Ultimately, O’Neill says, she’s much more concerned about this than she is about the possibility of AI sentience: “We cannot really have the conversation about harm and scale if we're also allowing the conversation to be distorted by thinking about AI being conscious when there's no meaningful evidence that there truly has emergent properties of consciousness in any of the models at this point. You know, could there be. That's open to conversation. But it's a ‘three drinks at a cocktail party’ conversation in Silicon Valley — it doesn’t belong being the basis of governance and policy decisions.” — Screencap from linked article: Ultimately, O’Neill says, she’s much more concerned about this than she is about the possibility of AI sentience: “We cannot really have the conversation about harm and scale if we're also allowing the conversation to be distorted by thinking about AI being conscious when there's no meaningful evidence that there truly has emergent properties of consciousness in any of the models at this point. You know, could there be. That's open to conversation. But it's a ‘three drinks at a cocktail party’ conversation in Silicon Valley — it doesn’t belong being the basis of governance and policy decisions.”

Prof. Emily M. Bender(she/her)

@emilymbender@dair-community.social · yesterday

My only quibble is that I am (again) paraphrased as if I talked about "AI" as a thing, or used "AI" to refer to language models. I'm sure what I said to Holly Baxter here was "language models" have these uses. I've asked for a correction.

In general, if you see me quoted/paraphrased in the media and the term "AI" is outside the quotes, that's gonna be a journalist mis-paraphrasing me.

/fin

Screencap from linked article:

In Bender’s view, there is simply no good way to interact with an LLM. AI itself can be “extremely useful for things like automatic transcription and machine translation. There's a role for that kind of technology,” she says. “But turning it around into this chatbot interface and producing synthetic text by repeatedly answering ‘what's a likely next word’? That is not technology that I see beneficial use cases for.” — Screencap from linked article: In Bender’s view, there is simply no good way to interact with an LLM. AI itself can be “extremely useful for things like automatic transcription and machine translation. There's a role for that kind of technology,” she says. “But turning it around into this chatbot interface and producing synthetic text by repeatedly answering ‘what's a likely next word’? That is not technology that I see beneficial use cases for.”

2+ more replies