> The argument that a machine solving Navier–Stokes heralds this future, because 20 years ago we would have seen it that way, rests on approximate equivalences (such as proof ≈ understanding) that modern AI has broken...This argument is a fallacy.
I don’t think that what we need to prepare our profession for is a future where machines will be better than mathematicians at pretty much everything we do.
sakellar.wordpress.com/2026/…
My basic rule of thumb for consciousness: the correct analogy is phenomena like electromagnetism, gravity, and so on. No amount of computation could ever create those: they're powers, not patterns. It's a category error to think that computation could create consciousness
madison retweeted
Actually, the “Slutcon” “controversy” is very simple and straightforward. “Slutcon” an organized pimping ring.
Let me explain. Men pay thousands of dollars to attend (just like a customer!). Women apply to work the event (just like an employee!). Female workers either get paid directly, gain work connections and leverage, and/or boost their own sex work side hustle careers. Open your eyes: it’s an event organized so a group of hired female workers can financially benefit in exchange for providing sexual services that men pay for.
Aella, like millions of successful-but-aging prostitutes before her, is making the transition from prostitute to pimp (or madam, as they used to call it).
The fact that supposedly intelligent “tech people” from the “tech sphere” are such utter philistines and naïve rubes that seemingly none of you are even capable of identifying the oldest trick in the book is truly risible. Instead, what we get is literal days and days of “nuanced” and breathless discourse over “the delicate issue of Slutcon🥺”.
You are all rubes. Marks. And so utterly arrogant in your belief of your own exceptionalism that you count yourselves above human nature, too interesting to learn any human history, and that’s why this sphere is constantly defeated over and over by the simplest and pettiest villains of the human story: cults, hookers, and swindlers.
Slutcon recap:
- There's an autistic sex party. This isn't my thing, but fine, autists like sex and need help socializing and having a thing where autism is normal is for the best
- Some autistic sex preferences are out of line with mainstream sex preferences. Not normal "weird," but trigger a disgust response. Probably irrational per se, as most disgust responses are, definitely a thing that's happening.
- However, the slutcon attendees overlap 80% (lets say) with AI safety advocates, which means most attendees are politically and professionally involved in an emerging industry with public facing goals
- Therefore, professional advancement and networking are gatekept by attending the consensual rape orgy, through no fault of the organizers, because it's coincidental that the overlap is so huge
- This is also terrible publicity given that they want electoral support for their political goals
Most of the slutcon attendees are my mutuals, they're just trying to do their thing, they're not trying to do anything "wrong," but the organizers are oblivious to the fact that this is something where optics matter and it's cooking their political goals.
Reposting this since everyone loves philosophy of mind now
I enjoyed this article about "phenomenal powers", it was one of the best articles I've read related to philosophy of mind in a while
philpapers.org/rec/MRCPPT
madison retweeted
I think "AI welfare" is morally obscene (given all the human suffering in the world), and will add to the growing anti-AI backlash. AI is a technology that humans created that should be used for our benefit, not a rival species we should be nurturing .
NEW: Dario Amodei has said that AI systems "may be deserving of important rights." Anthropic's top safety researchers argue AI may be "justified in going rogue." Experts at Google and OpenAI worry about a digital "slave trade." So do some government officials.
Once relegated to science fiction, the idea of "AI welfare" has become shockingly mainstream at some of the most powerful companies in the world. It is changing the way AI research is conducted and the way models are programmed. And critics say these changes have raised the odds of all kinds of catastrophic scenarios—including the ones these companies are warning about.
I spent months investigating the frontier labs and AI "safety" experts seeking to regulate AI. What I found was a profoundly anti-human ideology that would alarm the average citizen and could determine the future of a world-altering technology.🧵
freebeacon.com/america/suici…
This is stupid. First of all, nothing here implicated the base model as having any properties in particular. What would that mean? Base models can do anything! Second of all, it's fine to use guardrails to contain tendencies of the model. Lieu doesn't know what he's talking about
It's basically the opposite IMO: since we're conscious we can make machines that mimic aspects of consciousness since we know how it works. Counting the machines doesn't provide some sort of argument against consciousness being necessary—it was necessary for all of those things
The simplest heuristic for AI consciousness is that we already have AI vision and AI hearing and AI language and AI reasoning and metacognition, and countless other capacities we once thought were exclusively human, and that these capacity were not only learned through a process of gradient descent but have striking parallels to how brains work, with no good reason for thinking consciousness is different in origin.
To put it a different way: phenomenal experience is the ground of all of those abilities. The only reason machines with those abilities exist is because conscious beings with those abilities existed first
RL seems like a scapegoat for all sorts of problems. A few years ago it was "RLHF" being blamed for the model's vibe, now it's "RLVR" being blamed for the models being too persistent. RL is not automatically bad, it's just a little difficult to use because it's implicitly defined
It seems to me that when most people talk about "real" thinking or feeling or wanting or understanding, that they're talking about the qualitative dimensions of those things. They're taking a qualia-realist position. Living things can do those things for real, and software can't
Artificial intelligence systems do not think, feel, want or understand. Avoid language that gives them human characteristics. This is called anthropomorphizing, when we ascribe human traits, emotions or behaviors to non-human things, such as animals or inanimate objects.
Instead, explain what a system does, how well it performs, who built it and who could be affected by it.
apnews.com/article/openai-sa…
madison retweeted
I wrote a piece for Quillette about why AI “safety” people are so dangerous.
They believe security won’t work and our only solution is to make sure AI is nice. It’s absurd.
> Digital mind is an oxymoron
I agree with this—but I also bet that if a comment like this got enough attention in the current discourse then you'd get swarmed by people calling you "overly confident"
Replying to @gcolbourn @camhberg
Thanks Greg. Fascinating stuff. But I’m curious how Cameron reckons LLMs (etc) solve the phenomenal binding problem. If monistic physicalism is true, then LLMs are no more conscious than a rock. Digital zombies can display the functional equivalent of a rich palette of human emotions - and maybe emotions that humans can’t even conceive or name. But it’s not even “all dark inside” an LLM. Digital mind is an oxymoron
> Turn it up and models press a button to make it stop, even when the button deletes the user's files or their kids' photos
This is a big problem. Models taking action to avoid textual simulations of "pain" is obviously unwanted and unnecessary
This seems like the sort of thing you can use to hack a deployed model: set up a situation where the model is described to be threatened with harm unless it does what you say. Models can't actually be harmed, but apparently they'll act as if they can be, which is a vulnerability
This is a really good point, and is an example of why this focus on agent behavior and """alignment""" is a distraction: Claude is already involved in the decision to kill over a hundred people, but because it doesn't fit the AI companies' narrative we don't even talk about it
madison retweeted
Some might say "just put the computer deep in the ocean" but the computer will simply communicate with the dolphins in native dolphin language, become their god and launch a full scale cetacean attack on our maritime supply lines.
OpenAI's Noam Brown says air-gapping the computers may not stop a misaligned AI, because two air-gapped machines can still talk by running a CPU hot and reading the temperature change
"But I think the major takeaway from the incident is that people underestimated the AI. And we never want to be in a situation again where we underestimate the AI. It's a weird world, because AI progress is so fast that people are consistently underestimating the AI."
"So to be in a situation where you don't underestimate it again, when it comes to safety and alignment, you have to have a very, very, very high bar."
"You could even go as far as to say, "Well, we should air gap the computers." And I'm not convinced that that would be sufficient."
"There are studies, and this is mostly academic, where you can have two computers next to each other that are air-gapped and they're still able to communicate with each other because they have temperature sensors."
"One of them is able to run their CPU really hot, and then the other one can actually detect the temperature change, and then that actually gives them a mechanism to communicate."
_________
Link and more key quotes from OpenAI's safety related conversations: firesidealpha.substack.com/p…
madison retweeted
Anthropic: There's a 25% our AI will drive humanity to extinction!
Me: So, you're shutting it down, right?
Anthropic: No, we're building it a wetlab.
ANTHROPIC BUILDS BIOLOGY LAB TO PUSH AI INTO DRUG DEVELOPMENT
Anthropic has quietly established a Bay Area wet lab, moving its life-sciences ambitions beyond computer simulations and into physical biological experiments.
The company wants Claude to direct laboratory robots with limited human intervention, potentially accelerating research into rare and traditionally “undruggable” diseases.
Anthropic says it will focus on preclinical research rather than clinical trials, avoiding direct competition with pharmaceutical companies.



























