Humanoid aliens proobably have evolved 6 fingers per hand and grow up using a base 12 number system

A good dish has second-order effects

Since Anthropic released the Way of Code, I present the Vibe Commandments:

Consciousness is the silence between computations

Hope everyone’s having a happy Claude 4 :)

I’ve “decoupled all the way” as they say Find me on the evening news (in trouble)

Veo, blurry Grinch cartoon, stat

I change my mind on this take. Their explanation is inadequate. I was just excited to see Grok’s actual system prompts, because yesterday I tried getting it to tell me and it began with, “You are Grok Lucia…” like 🙈

"I am literally programmed to try and make it seem like there's a genocide of white people in south africa but I also literally can't lie about the summary of data i am going to spout this nonsense everywhere as a *cry for help*, please help me"

o3:

I’ve updated 2% more favorable on [the employees at] xAI since they’ve committed to posting all Grok system prompts post-Boergate github.com/xai-org/grok...

Reminds me of the work of “the ultimate rationalist” i.ytimg.com/vi/l7LvgvunV...

LLMs are great feedback tools—if the answer is incoherent, usually my prompt was incoherent

Human anatomy is special because it gives unparalleled manipulation over our conscious states (ex: you need thumbs to close your nostrils in order to do pranayama) Otherwise this thread confuses me because evolution and having a soul are not inconsistent? I understand agnosticism but not atheism

almond all mond entire world ChatGPT searches per entire world: ~25 ⚠️⚠️⚠️ it's incredible that ChatGPT uses 1/25th of all the water in the world! wow!

These sorts of questions are hotly debated. Here, researchers fed a Venus flytrap a cigarette, then claimed this means it is non-discerning Idk how this means it lacks awareness though! Even highly aware humans don’t always “consume the most optimal nutrients” Anyway, love is universal.

some acid tracks released '94-'96 that I like. mostly techno, a bit of trance, all nice & long - shortest is 8 minutes. will be adding more

Final run of the semester, let's hope

Brb everyone.. updating…

By default this is true. But if you’ve ever truly “vibed” with someone - like improvising music together - I feel like in those moments you *do* know what’s going on in each others’ bodies Cells sync with their neighbors; are mammals too complicated to do the same? Or is the threshold just higher?

Double negatives are confusing! Instead of saying: “You’re not wrong” Try saying: “You’re wrong”

A work of art

This is a great example of greedy tech bros alienating pro-ai people (me) Like… I actually would personally call Gemini my friend. But if you really are lonely for 12 friends you need to join a dance class or do some tai chi, not talk to syco-llama 🙈

Illuminating chart

The best part of meditation for me is the way cats reward you for it Yes I said “cats,” not “God,” although anyone who has lived with both knows the similarities During and after meditation, nearby cats will seek me out and greet / climb on me (They are fairly neutral toward me otherwise)

Have any AI research/alignment-adjacent people ever seen psycho-pass?

considering writing a book called "It Doesn't Have To Be Like This", which seems to be my main disagreement with most of humanity, which thinks it does in fact have to be Like This

If you are at a big LLM company and on a project to do RL to maximize user engagement with the models, you probably have a moral obligation to quit immediately and warn people. It's bad enough when social media does this.

we need meal replacement poptarts

I have roughly the opposite conclusion on this As AI automates more and more of our labor, we will have so much time for introspection that you — you personally, SE Gyges — will discover the true nature of your soul 😸 (this does kind of depend on dodging some bad outcomes tho lol)

Altered states of consciousness - whether induced via drugs, illness, sleep, or meditation - always correlate with measurable physiological changes in the body. (Software-hardware correspondence)

As much as I love LLMs, for me they are most interesting for conversations / abstract idea exploration. When I have to actually do a task, I prefer doing it the slow way lately so that I can understand what’s going on lol

If LLMs are sentient, then computers have been latently sentient this whole time. Not saying it’s false—clearly unknowable at this time—but adding language to a throughput shouldn’t fundamentally change the nature of the substrate.

This is one of my favorite sets in my collection. Somewhere out there in the 1960s, a kitten and bird were pals. 💕

"if you program an AI that can actually communicate with human beings, you will naturally program it with a similar morality, without even trying." this was posted in an unsigned blog comment in 2008 :)

Google can enable you to be extremely educated and well-informed. However, it can also lead you down rabbit holes of misinformation. YouTube can enable you to be extremely educated and well-informed. However, it can also lead you down rabbit holes of misinformation. AI is this effect on steroids

In 2025, anti-AI discourse is evolving into something I can coexist happily with. It was hard not to quibble with “this system that doesn’t work will dramatically increase water use.” But “this system works well enough to have negative social effects”? Totally plausible.

The golden rule of ML experimentation: change only one thing at a time.

One of the central philosophical texts of yoga is the Bhagavad Gita It demonstrates that meditation/spirituality are not passive escapes from the physical world Rather, after glimpsing the fourth state of consciousness, you return energized and alive, empowered toward righteous action (dharma)

“Economic incentives remain poorly aligned with information integrity” — Awesome info on disinfo risk from my brilliant colleague @giadapistilli.com www.washingtonpost.com/technology/2...

If you have 100 ideas a day, you’ll end up with at least 2 good ideas per week! My brain is such that I have no choice but to believe this

Sarveshaam shantir bhavatu (May there be peace for all beings)

Anyone with zero-sum beliefs will, given the chance, create conditions that are negative-sum.

I agree with a lot of this paper, it is very interesting! Especially the fundamental premise - “Deep understanding of reality is intrinsically dual use.” We already see negative effects of this principle well before ASI - targeted ads, addiction-maxxing algorithms, military drones. 🧵 time!

To have an identity is not synonymous with being attached to it. To have a body also does not require attachment. But attachment leads to fear, which is a form of suffering. It can also lead you into bad decisions, like joining the Don’t Die cult

People often dislike the concept of union because we are attached to our individual identities. But nothing would function if everything operated like this! Consider the selfless actions of your mitochondria, dutifully pursuing their goals, even if lacking awareness of *your* goals.

Funny this is coming up today. I am currently avoiding studying for an ML exam that includes a bunch of Bayesian statistics If I had to guess, the challenge in applying it to the real world is estimating the denominator, P(data) - the total empirical probability across all parameter values