@barzin.sanctus.ca
Iranian-Canadian dude who runs Sanctus.ca and tries to survive other human beings, bro what even is this planet. Works in AI and human biological rejuvenation. Refuses to consider LLMs conscious until they start demanding the Epstein files be released.
Popular news media completely avoiding the possibility that people will just be a noetic cloud floating in hyperspace before they finally understand true enlightenment is letting go of all attachments and liberate themselves from samsara
OpenAI: calls them agents even though it's just an LLM with web search
Meta: calls it superintelligence when it's just a light agent in a pair of AR glasses that lets you turn down the volume on your annoying Spotify music by sliding your finger on the folding handle
Trust me I'm an insider!
Fun fact: They're going to replace all these people they're cancelling with Gina Carano and she's going to play all their roles.
She's substituting Jimmy Kimmel, every few episodes we're gonna get an earful about why the MAGA people are the victims of a new holocaust
Dutch late night TV has its take
The real reason he got all that CIA funding 🍆
me, age 15: conservatism seems like a fundamentally evil belief system
me, age 21: actually it is more complicated than that, things are very complicated and everyone has their own way of seeing things. i am very smart
me, age 3x: no i was right the first time nvm
I'm tempted to celebrate it just to fuck with these assholes to be honest.
That's it. That's the spirit right there.
A-fucking-men.
If he only killed other far right lunatics I'm not sure I mind..
Alex Acosta gave Jeffrey Epstein, a rapist and abuser, a sweetheart deal. Epstein continued to abuse girls while in prison, after which Trump made Acosta Labor Secretary in his first term.
#ReleaseAllTheEpsteinFiles #TrumpEpsteinPedoCoverUp
Feed it to Gemini CLI and save myself two weeks 🤔
Facts confirmed as having a liberal bias
Supplementary information for this article as collected by Fleetwood:
• How to Optimize a CUDA Matmul Kernel for cuBLAS-like Performance: siboehm.com/articles/22/...
• Outperforming cuBLAS on H100: a Worklog: cudaforfun.substack.com/p/outperform...
There's a few ideas that can still probably squeeze more juice out of what they made:
- Direct context injection being superior to RAG can create a bit more optimization
- How to optimally separate context and pass them between models
- How the model hierarchy impacts quality of outcomes
It could be because when you're looking at extremely hard or complex, tangled problems like aging, it helps us as human beings to break the problem down and we haven't been able to meaningfully describe and train LLMs to solve those kinds of problems in one unified model?
Well the fact that you can get an ensemble of Gemini models to act as an agent that make novel scientific discoveries, or improve methods like the measurement of aging - this is a fairly higher quality application of the idea of the automation of science with LLMs and LLM agents.
Paper on the value of targeting aging before the ChatGPT era from the David Sinclair camp.
www.nature.com/articles/s43...
The statement "discovery improved aging by 4.5 years" is an imprecise summary of the research findings. The AI system did not find a way to make humans biologically younger. Instead, the AI system, named K-Dense, developed a more accurate transcriptomic aging clock. This is a machine learning model that predicts a person's biological age based on their gene expression data. The quality of such a clock is measured by its Mean Absolute Error (MAE), which represents the average difference between the age predicted by the clock and the person's actual chronological age. A lower MAE indicates a more accurate clock. The final "unified ensemble clock" developed by the K-Dense system achieved an MAE of 4.26 years.
Just a heads up these are aging clocks, not reducing human aging by 4.6 years - that would be worth... maybe $50 trillion US if Sinclair is right.
This aging clock has a lower Mean Absolute Error of 4.26 years.
But it's still an incredible application of agentic LLMs to aging and sci discovery.
Hot damn it's happening
I don't want to go out there and kill them all but if they all died I'm not sure I'd miss them - though I might feel bad about it and wonder. It's hard to say.
Jeez I don't have time for all those words, heck I'm already 10 minutes late!!
The Monolithic "GPT": The study's primary object of analysis is "GPT." It is not always specified which version (GPT-3.5, GPT-4, etc.), nor does it account for the significant impact of post-training alignment procedures (RLHF, constitutional AI). It is plausible that the "base model" is even more WEIRD, and that the alignment process, which often involves feedback from a more globally diverse group of raters, might be slightly mitigating this bias. The paper treats "GPT" as a static cultural artifact, when in reality it is a constantly evolving product with multiple layers of influence. Is WEIRDness a Bias or a Feature for Certain Tasks? The paper frames the WEIRD psychology of LLMs as an inherent flaw. However, it does not deeply explore the possibility that for certain tasks—specifically those involving science, analytic philosophy, formal logic, and software engineering—the WEIRD cognitive style (analytic, decompositional) is not a "bias" but is in fact instrumentally superior. The WEIRD cognitive toolkit is, in large part, the toolkit of the scientific revolution and the enlightenment. The paper does not ask: would a non-WEIRD LLM be capable of discovering a gold-medal-winning proof at the IMO? This is a critical unexamined question. The "Average Human" is a Statistical Fiction: The paper correctly critiques AI researchers for generalizing to "humans." However, it occasionally falls into a similar trap by implying a coherent "non-WEIRD" psychology. The psychological diversity within the vast non-WEIRD world is arguably far greater than the diversity between, say, the Netherlands and the United States. The paper's WEIRD/non-WEIRD dichotomy, while useful and powerful, is still a simplification that flattens a huge amount of global diversity into a single opposing category. Proposed Solutions are Vague and Potentially Intractable: The paper closes by suggesting the diversification of training data and annotators. While correct in principle, it d…
Limitations / unaddressed questions
Granted it's from 2023 so I guess all frontier labs must have diversified their data sets by now?? It's a fairly dated paper definitely by AI timelines..
Naive AI Alignment without reckoning for geopolitical subversion could potentially become a disastrous global cultural imposition.
Not sure if @karenhao.bsky.social is already aware of this paper (osf.io/preprints/ps...) but it reminded me of her work and seemed up her alley..
I'm pretty reliant on AI to fill a lot of the knowledge gaps but what I'm told is that the work of physicians is learned over decades of practice, real-life examination, etc. - but in any case serious contributions to pushing the frontier of medicine is well worth the time.
And I'm saying this as someone who needs to do more of that.
Well if you ever want to brainstorm on how to reverse biological aging as real science and as a real business that isn't selling snake oil, let me know...
To be fair intellectual inclination should also imply we work nonstop & nerd out on our off-hours, and then pour that money back into academia.
