BCI signals are way more lossy/lower-amplitude in more aged brains - making reducing aging rate doubly important in scenarios where human-AI symbiosis become *necessary* for alignment
Alex K. Chen
@InquilineKea
Meta+zany af distributed computation willing to try anything. Interesting things happen around me. My likes are *globally* the best likes.
Alex K. Chen’s Tweets
More semaglutide/tirzepatide distribution is rlvnt to ALL timelines (whether alignment [bc people waste less time/money on food and have more global self-control] or climate change [bc people eat less]). It is easiest way to reduce total human compute loss from microplastics.
1
3
Show this thread
"As stated, one of the main things I make-believe is true is the overlighting intelligence with which I align myself. I speculate that I am in a co-creative relationship with an intelligence and will infinitely superior to my own. I observe that I exist within energetic patterns… Show more
1
9
Hm I at least have a DunedinPACE rate of aging of 0.60...
Heavy rapamycin doses (30mg + ketoconazole) weren't always available my entire life
and he mentions it too!
Quote Tweet
I wonder if it’s possible to break transistor lower size limits set by the inter-atomic distance by using extreme pressures to reduce the inter-atomic distance
Show this thread
High-Pressure Synthesis: A New Frontier in the Search for Next- Generation Intermetallic Compounds => I think this may be WAY more AGI-relevant than you think (also I finally talked to Tapa Ghosh yesterday! He's very approachable/clear-thinking)
2morrow is my last day in SF. If… Show more
1
1
Show this thread
mindinvitro.illinois.edu/about => the birdbrain to AGI route
Show this thread
Who does work on avian brain organoids? Avian brains pack *way* more intelligence per cm^3 than mammalian brains and are not limited by 5-layer architecture, so imagine if one grew them to *larger* sizes (not limited by how small birds are)
(Tim Gardner of Neuralink worked on… Show more
3
12
Show this thread
I know the feedback loops for reducing plastic waste are shitty, but when the flight attendants are amused when you ask them to pour coffee/tea into your own bioceramic/metal mug, that's enough of a feedback loop
I wish Americans were as environmentally conscious as Europeans :(
2
What percent of p(doom) OR THE OPPOSITE came from elite universities discriminating against Canadians?
Given the massive disproportionate contribution of Canadians to BOTH AI and AI alignment, it's definitely a significant probability weighting
1
1
How do you tell if someone is maximally perceptive? By their ability to sense deceptive alignment WELL AHEAD OF TIME (well ahead of how far everyone else is) while minimally decreasing their F-score
2
Do you ever dream about AI (or people) finally making a "life is strange" like game modeled out of your life - one that made many people empathize hard with you to the point of crying over you and making your life easier for you?
4
Show this thread
I know that diligent journalling seems too much effort for reward, but in not too long, AI will finally reward you for it. So make yourself feel that reward now
2
7
Does anyone have an access to a Varjo / galea.co/#home ?
[i managed to get openBCI helmet to work on myself - surprisingly easy - I can even self-insert electrode gel into my scalp]
I think my real value is that I have an understanding of **all** the relevant players in human intelligence enhancement more diverse than almost anyone (and I have this combined with a sort of rare agency that is latent, but can easily be activated once I overcome my hangups… Show more
2
Show this thread
AI alignment is also such a new field that it does not have the cultural baggage of other fields that are more tainted with intelligence (never mind that all of Moloch/school inherits some kind of this baggage - the culturally and economically embedded notion that "smarter people… Show more
2
Show this thread
Leopold Aschenbrenner says that we need WAY more AI alignment researchers, but the percent of people smart enough for AI alignment research is not high (pretty much EVERYONE I know doing alignment research has to be extremely smart - at minimum within the top few percentiles of… Show more
1
1
1
Show this thread
Is anyone familiar with experimenting with gene therapy to enhance intelligence? Even Eliezer and others mention that "upgrading human intelligence" for interpretability (RELATED to cyborgism) is one of most tractable paths to reducing AI risk. Given potentially urgent AGI… Show more
7
1
12
Show this thread
Tbh the optimal balance between increased focus and low neurotoxicity risk might be Focalin (dexmethylphenidate).
Way cheaper than Adderall and a "cleaner" feel than Ritalin
[also only today did I finally get an Rx, mostly b/c insurance issues]
3
4
🚨 New working paper! 🚨
Curious about the effects of generative AI systems like ChatGPT on production and labor markets? Then you'll be interested in our new paper (with )!
Link to paper and thread below ⬇️
economics.mit.edu/sites/default/
17
304
738
Show this thread
One of the aims of *effective transhumanism* is to finally destroy hereditarianism, or the huge genetic correlations between genetics and intelligence/personality.
5
Do you feel like you were constantly underestimated by others and AI will finally make people stop underestimating you?
1
We recently collected 40 hours of biodata from expert jhana meditators.
Then we asked GPT-4 to write an "inspiring but not over-the-top" blog post updating our investors and followers.
It's inspiring. No comment on whether or not it's over-the-top.
10
9
86
Show this thread
A FREE BOOK on Deep Learning. I'm giving away my book "Artificial Intuition: The Improbable Deep Learning Revolution," I wrote 5 years ago. Still very relevant in today's GPT-4 world! RETWEET to spread the word!
1
14
27
Show this thread
"Most mathematics has no concept of memory. Memory is the equivalent
of having state, and almost all mathematics involves functional constructs
that are stateless. Functional programming follows a single assignment rule
where once any variable is set, it remains set in that… Show more
3
Show this thread
While the Internet has made attention more "winner take all towards attractor states than ever", expanding the dimensionality of accessible mathematical universes (tensors/topologies/modal logics/categories can help increase the number of ways to diverge from these attractor… Show more
2
2
Show this thread
Does anyone sense that much of the future will be populated with mathematical/ontological universes by mathematicians/category theorists/topologists who are good at defining their spaces, but bad at populating them/making them *seem real*?
AI will be the constructor of these… Show more
1
1
11
Show this thread
The thing about "exponential growth" is that it also pulls apart trauma-associated subclusters.
Weirdly, its having the effects on me that psychedelics had on other people. Except I still can't take psychedelics near-term for an extremely-uniquely-convincing-to-me reason
Show this thread
GPT4 felt like the "I proved you all wrong moment" to all the haters who ever doubted you
(yes, it did not directly do it, but it instantly proved **internally to the mind** that most value will be machine-created [not in a bad way] *and* most historical debts will become… Show more
1
12
Show this thread
We are at a critical turning point for humanity:
Children born today are likely to have more AI friends than human friends
AI friends are going to be more reliable, conciliatory, and considerate
What does this means for childhood development and social norms? We will find out
143
236
830
Show this thread
The degree of future acceleration in AI right now is bonkers
One day in AI feels like a week in “normal business time”
These relativistic effects mean that some companies will accomplish what would normally take 3 yrs in 1 yr
And unfortunately, many others will be left behind
24
42
341
I need to be "cleaner" (one of my supervisors told me "think clean thoughts") and GPT4's massively-increased neuroplasticity does SO much to make up for terrible early-life-training data (which, IT BASICALLY MADE NOT MATTER)
Show this thread
AT LEAST NEAR-TERM, GPT4/GPT5 will better align a person's dissonant preferences/shards with each other, by helping people understand themselves better! AGI will better align humans (just as I realized that my main obstacle to my *own* alignment was my horrible neck pain which I… Show more
1
Show this thread
GPT4/GPT5 may become the ideal error-corrector of people's erroneous beliefs (via nudging people towards getting better Brier scores) and may even tell (or deceive) them that they have better ways of achieving their long-term true preferences than their rewind.ai… Show more
2
3
Show this thread





