iiinova — where your digital self keeps evolving

16915

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. >>

31698

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: dl.acm.org/doi/10.1145/... >>

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].

317821

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. >>

222619

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: >>

216210

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. >>

126335

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. >>

528435

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. >>

525830

The Distributed AI Research (DAIR) Institute @dairinstitute.bsky.social · 3h

Anthropic is out with new nonsense about Claude (or the collection of models called Claude) deserving moral consideration and/or be 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: >>

11399150

Reposted by Emily M. Bender

AI doomerism isn't sound science, it's a parlor game. Great to see reporting that pushes back and quotes experts like @timnitgebru.blacksky.app and @emilymbender.bsky.social. Coxon and Anthropic "did not respond to a request for comment." 👀 👀 👀 @thebulletin.org thebulletin.org/2026/09/how-...

thebulletin.org

How the bad science of AI doomerism is good for big business

There is no clear path showing that today’s AI systems will lead to superintelligence, much less a basis for calculating the probability of doom.

14511

But also my memories of Dharamsala are bringing me hope and gratitude because it wasn't only that --- despite strong pressure from Silicon Valley, religious and contemplative leaders and practitioners aren't all or even mostly chasing after the shiny, futuristic fantasies.

0100

Emily M. Bender @emilymbender.bsky.social · 3h

Religious and contemplative traditions can be great sources of knowledge and wisdom on how to navigate our world and build strong relationships and communities and it hurts to see them focused on the artificial. >>

1100

And every time I pause to look back at that, it seems like the hype has gotten bigger, more extensive, and more insidious. This week, I am seeing credulous media coverage of Anthropic consulting with religious thinkers to try to work out how to "grow" Claude into a moral being. Ex: >>

nytimes.com

Religious Scholars Met With Anthropic. What They Heard Stunned Them.

In a series of private meetings, the company consulted religious scholars to help instill morality into its A.I. models — and make the case that Claude could be conscious.

1101

Emily M. Bender @emilymbender.bsky.social · 3h

Ever since Spring 2023, when I was teaching a seminar on "Meaning Making with Artificial Agents" and it seemed like every day the news was talking about ChatGPT, I have been looking forward to the day when the era of AI hype is in the rearview mirror. >>

2100

Emily M. Bender @emilymbender.bsky.social · 3h

A year ago today, I was on my way to Dharamsala, India (via a 24 hour stop-over in Amsterdam to teach my classes online) for the Mind & Life Dialogue, a truly wonderful event. buttondown.com/maiht3k/arch... >>

buttondown.com

Reflecting on "AI" in the Himalayan foothills

By Emily The Himalayan foothills, as seen from the kora walk around the Dalai Lama’s temple complex Last week, I had the immense good fortune to participate...

1211

Emily M. Bender @emilymbender.bsky.social · 5h

That said, it feels off/sad that we have to start by creating zones where people are expected NOT to use the pervert glasses, or Apple watch always-listening tech, etc etc. OTOH, let's start somewhere!

1241

Emily M. Bender @emilymbender.bsky.social · 5h

The notion of "second-hand surveillance" (on analogy to second-smoke) seems powerful -- and more generally I really appreciate the idea of zones where one can be free of this. From @ajlunited.bsky.social www.offrecordzones.org >>

offrecordzones.org

Off Record Zones - Places Where Consent is Respected

23310

Yes, you are confused. Perhaps reading the rest of the thread you are replying to and/or the actual linked article would help.

150

but the work of @kirbyconrod.bsky.social and others has helped me understand that some intentional language change is important and beneficial. /fin

0140

One of the very first lessons in Linguistics 101 is that descriptivism is our goal (studying how people talk, not how they should) and prescriptivism is bad (a kind of oppression/discrimination), >>

1190

Is it through linguistic discrimination and stigmatization of patterns of language use, or through education (here's how to write about xyz to better inform the public/here's how to refer to people to show respect)? >>

180

As with just about anything else, it depends on both why and how the pressure for change is applied. Is it in the service of propaganda and nationalism or in the service of clarity and/or respect? >>

1110

It's interesting to think through authoritarian pushes for language change, where the authoritarianism and the prescriptivism are both obviously bad, but then also reflect on cases where recommendations for deliberate language change can have positive impact. >>

1141

This is a pretty abbreviated summary of that part of our conversation, but I think it gets the key points: >>

Screencap from linked article:

Bender told me that when there’s a push for a change in language, linguists always find it important to ask what it’s for — whether it’s serving propaganda or nationalism, or whether it’s genuinely helping people understand better or show respect, for example, with certain names and pronouns.

“We’re interested in how people talk, not telling them how they should talk,” Bender said.

1210