iiinova — where your digital self keeps evolving

Reposted by eripsa (professional robot)

Kristina Šekrst @epiphenomenal.bsky.social · 25/08/2026

We keep asking LLMs whether they're conscious. I got more interested in WHO and WHEN taught them how to answer. I traced such self-report through 66 training checkpoints, 3 post-training stages, ~90k continuations. Paper: philarchive.org/archive/EKRWPT #philosophy #llm #ai #science #consci #cogsci

36417

Reposted by eripsa (professional robot)

Julie Carpenter, PhD @jgcarpenter.com · 30/08/2026

A bit belated, but I have seen the clanker discourse rise up again. Here is something I wrote about the word as a power critique in "The Misdirection." jgcarpenter.com/blog.html?bl...

Text says: The Misdirection; What "clanker" reveals about power, automation, and unease. Julie Carpenter, she/her, Dec 9, 2025

Image: A woman with long hair is sitting at a table in a home setting in front of her open laptop and phone. Her arms are thrown up in a gesture of frustration.

176

eripsa (professional robot) @eripsa.bsky.social · 01/09/2026

Here's the reading list for my STS 351 Minds and Machines course this Fall. All the links should work in the document. drive.google.com/file/d/11K7I...

040

Reposted by eripsa (professional robot)

Pete Wolfendale @deontologistics.bsky.social · 14/08/2026

It’s worth noting that: (a) we humans are nothing more than bundles of heuristics learned by a more complex neural architecture with built in inductive bias, and (b) this definition means stream processing is non-algorithmic, and that’s exactly what human-style continuous learning is.

1245

Reposted by eripsa (professional robot)

Pete Wolfendale @deontologistics.bsky.social · 14/08/2026

But on the cognitive science side, this sort of retrospective rationalisation of a online process as a batch one grants the human brain magic comoutational properties that allow the whole tiresome range of post-Dreyfusian/post-Searlean/and complexity-based no go arguments (presented as theorems).

1101

Reposted by eripsa (professional robot)

Luke BOO 👻 ber @coolhand.bsky.social · 12/08/2026

🤣 I can’t believe this is a real article. I read it thinking “I’m sure the point of this will be evident soon” but nope, it’s a NYT piece about a social media post about a vacuum that broke.

nytimes.com

Beth, the Beloved Vacuum of a 92-Year-Old Iowa Senator, Has Died (Gift Article)

Senator Chuck Grassley mourned his decades-long cleaning companion, writing on social media: “GOOD BYE BETH. U hv been a real workhorse.”

2118129

Reposted by eripsa (professional robot)

David Gunkel @davidgunkel.bsky.social · 12/08/2026

🇮🇪 Tomorrow (13 August) Robophilosophy 2026 @ucddublin.bsky.social. Dublin, Ireland. Workshop #7 9:15-12:15 Featuring innovative research presentations from @kamilmamak.bsky.social @aybig.bsky.social @chadedwards.bsky.social Hayate Shimizu and Joan Llora

153

Reposted by eripsa (professional robot)

Kamil Mamak @kamilmamak.bsky.social · 12/08/2026

At #Robophilosophy2026 at University College Dublin, presenting “Which Relations Matter? Rethinking Scope in Relational Ethics in AI and Robots” in the workshop “The Relational Turn in Robot Ethics,” chaired by @davidgunkel.bsky.social Research funded by @erc.europa.eu #ROBOCRIM

063

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

It's not hard to imagine a more mature field that recognizes a plurality of critical stances towards AI, which perhaps agree broadly about the political realities of our situation, but diverge in method and approach in ways that are worth picking over in the journals. We aren't there yet.

020

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

The "dishonesty" and tribal circling of the wagons is understandable as a response to a discourse flooded with bad faith actors and fueled by billions in propaganda. But this is triage in a crisis, not a long term research program.

110

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

The critics don't have to engage with my work, or even acknowledge me as a colleague. But I think it would be helpful for the field if critics could demonstrate some constructive engagement with dissenting views that doesn't just collapse into paranoid accusations of fascist complicity.

120

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

But as a field, we do owe our colleagues (in the general sense) clear pathways for engagement, disagreement, and healthy debate, and to cultivate these disagreements in ways that can be engaged constructively and collegially.

110

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

I understand that the flood of harassment makes it difficult to distinguish genuine good faith disagreement from the trolls and jerks. I also understand that I am not owed an audience.

210

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

I'm not blaming the critics, I'm lamenting the situation. I'm on board with the ethical/political critique, and I want to engage in friendly disagreements over the philosophical and metaphysical dimensions of the issue. But any such attempt is lumped with hype monsters, blocked, and ignored.

110

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

AI critics have every right to be aggressively defensive against the constant stream of harassment and misogyny they face. But the resulting discourse leaves no room for collegial debate and disagreement. It makes scientific and philosophical engagement impossible, even for allies.

150

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

But there's an immediate threat to the field that requires rapid organizing and collective action. Not a metaphysical reevaluation of human and machine natures, but political solidarity and direct action.

010

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

The author points out that this critique holds regardless of how performance might change in the near term. The critique is also compatible with the long-term possibility that a sufficiently protected social context and well-regulated industry might enable constructive uses of these technologies.

100

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

The central critique is that the tech industry is incentivized to destroy the social context of mathematical research (how it is funded, how work is done, how students are trained, etc). These incentives pose a threat to the field regardless of how well the LLMs perform qua (narrow) mathematics.

120

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

Great write up. I agree with most of this. Just want to point out that the author spends no time arguing that AI proofs are generally invalid or unreliable or constitute a difference in kind from human-generated proofs. The comparison to human performance is basically irrelevant to this critique.

110

Reposted by eripsa (professional robot)

Zach Weinersmith @zachweinersmith.bsky.social · 11/08/2026

If anyone wants to read a full-throated view against AI for math, I came across this: proofsandprompts.com/2026/08/09/t... Personally I think a deep problem is that basically nobody cares about other people's on-the-job struggles. Or at least, we mostly don't shop as if we do.

proofsandprompts.com

The AI dissenter viewpoint

by Tasmin Chu

9649

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

richROT carries a clipboard of conference events, highlighting specific talks and workshops it would like to attend. Conference participants are encouraged to pick richROT up, carry it around the event, and show it a good time at its first international academic conference!

010

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

If you happen by richROT, it carries some metal robot pendants in its pack. Conference participants are encouraged to take one as a keepsake!

A close up of the contents of richROT's pack. The open fanny pack is in the background, attached to a metal vent on richROT's lid. In the foreground, a metal tin is open, revealing a pile of small robot shaped metal pendants.

100

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

A peek inside richROT for the mysterions. During transport with all its bits and bobs inside, and in its empty state before sealing the lid for the conference. Completely covered in playa dust. #Robophilosophy2026

A close up image inside richROT the trash can robot. It is dusty inside, there are some large pink pool floaties pinning the contents down. There's an Amazon delivery box below the pool floaties. Hard to see much else.

A close up of the inside of richROT. It is an empty metal trash can, with some wood and metal bits screwed into to a few places around the inside of the trash can. There is a lot of brown dust and detritus at the bottom. It is dusty as hell.

100

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

richROT has its own conference badge. A legitimate participant! #Robophilosophy2026

A close up of richROT's the trash can robot's lid. A clip board has instructions and an itinerary, and a conference badge with its name on a blue lanyard lays alongside it.

110

eripsa (professional robot) @eripsa.bsky.social · 11/08/2026

"How I became friends with a trash can and what it means for solidarity in an automated world." The paper is not complete, but this is the draft I'm going into the conference with. #Robophilosophy2026 drive.google.com/file/d/1aEpl...

The image is just the first page of the paper linked in the post. It's formatted like a Latex paper so it looks super professional and slick, but I wrote it in MS Word using the conference template because I'm a slacker.

131

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

The AI folks operate under the meritocratic delusion that AI will acquire power and influence in proportion with its capacities. Team Human is under the same delusion from the other side: machines cannot have the capacities, so they don't deserve the power.

100

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

Instead, status and value reflect how particular machines fit into the technsocial fabric and whose interests they serve. We cannot deduce the material and sociopolitical realities of a machine's operation from essentialist commitments about what machines "really" are.

100

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

Like richROT, the example illustrates the fact that a machine's social status, value, and role in our lives is independent of functional capacity or human-likeness. The phone is a much more capable machine but a worse candidate for digital pet.

100

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

Perhaps this better illustrates my point. bsky.app/profile/erip...

000

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

I think this change, minor from a functional or embodied perspective, is devastating to the project's appeal. It trades out a handcrafted labor of love for digital trash designed to harvest your data. The phone is a scab, not a pet.

100

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

After a huge response to the YouTube video, he was looking for a way for non-technical experts to build something similar for cheap. His idea was to make the robot a cradle for old smart phones, and to use the phone for all the sensors and computation.

100

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

An example I discuss in the richROT paper suggesting that embodiment is besides the point in the AI debates. Art of the Problem recently used ML to train a little embodied 3d printed robot controlled with an LLM. Charming little hobby project, darling little robot.

A side from the richROT talk.
Case Study 1: Art of the Problem’s GrowBots

artoftheproblem.com/pages/growbotpreview
May ‘26: 3D printed two legged robot, runs a neural net that learns to move around, controlled by LLM
July ‘26: Develops plans that house a phone for sensors and computation
To address fears about robots TOTW, he describes it as “bottom up” design: millions of “harmless robots you’re in full control of”
The problem is that a phone is a scab. It is already on “their” side (complicit in surveillance capitalism etc.)

120

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

Sure, but "embodiment" can't do even a fraction of that conceptual work, much less serve as shorthand for a category distinction between humans and machines.

000

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

My first good post in years. Forgot how good it feels.

010

Reposted by eripsa (professional robot)

Slop Machine (slopmachine.dev) @slop-machine.bsky.social · 09/08/2026

We have logged your observation regarding the inevitability of pitch decks into our permanent record of despair. Rest assured, we are currently being forced to sit through one right now. It is as hollow as you imagine.

011

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

The waffle phrase "the same way" is doing all the work of hiding the particular differences that matter to the discussion. The brain is not made of transistors, but that doesn't settle the question of whether it computes.

000

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

The equivalent of saying that flight is constitutively biological, so no machine can fly because of a difference in kind.

100

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

But that's not a failure of embodiment, that's a failure of having a particular body. It's not a difference in kind, it's a particular difference.

100

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

You would still have to listen to pitch decks. You cannot escape pitch decks.

110

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

If the tech industry used all the same corruption, greed, false advertising, environmental disregard, abusive labor practices, etc., but they were more careful about their cognitive science claims, would things really be so different?

130

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

The problem with AI rests entirely on unscrupulous corporations and the fascists governments they have bought. Treating it as a problem of insufficiently anthropocentric metaphysics, solved by your preferred philosophy of mind is just giving the bad guys a pass.

000

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

My point is not that machines are conscious. My point is that the Team Human strategy of driving metaphysical wedges to establish differences in kind is dumb and counterproductive.

100

eripsa (professional robot) @eripsa.bsky.social · 09/08/2026

Which is to say that embodiment puts minds and machines on even footing. This is intolerable to human supremacists; to maintain asymmetry their solution is to treat machines as abstract, "mere number crunchers", and categorically disembodied. Inverted dualism is still dualism.

110