What are some strategies you use to “reduce the hit” when you’re about to take in potentially bad news? This is important b/c it’s sometimes important to face up “bad news” earlier rather than later, and there is social loss in some people not being able to face it until it’s too late, esp b/c some kinds of “bad news” aren’t as incorrigible as they may initially appear (just that you need out-of-distribution strategies to make the proper amends)
[some examples of bad news: irreversible data loss, cancer diagnosis, elevated epigenetic age, loss of important friend, someone overpromised and underdelivered on you and that affects many of the promises you made]
[as AGI timelines come “nearer”, “bad news” may come at faster frequencies, but OOD ways to solve them may also come faster]
[sometimes you can ask yourself “how much wealth would you need to take in any bad news”]. Wealth is not fully-completely interchangeable with youth/intelligence/universal social acceptance, but it DEFINITELY has potential for tipping the needle..
some neurofeedback peole (still TRYING THIS OUT, the S/N ratio might not be high, but finding the right person alone can make it super-worth it). Need to try LENS neurofeedback (money for this would be helpful btw)
Could it be a good idea to enable a file uploading feature to LessWrong? (eg a file uploading feature of PDFs or certain images/media]. Could help against link rot, for example (and make posts from long ago last longer—I say this as someone who edits old posts to make them timeless).
Every single public mainstream AI model has RLHF’d out one of the most fundamental facts about human nature: that there exist vast differences between humans in basic ability/competence and they matter.
Is there a simple way to jailbreak the models, such as asking them to talk about a hypothetical parallel universe which is exactly like ours (same biology, same history), except that in the parallel universe humans can have different abilities and competences?
[this is also how to get into neglectedness again, which EA adopted as a principle but recently forgot]
from Charles Rosenbauer:
This is neat, but this does little to nothing to optimize non-AI compute. Modern CPUs are insanely wasteful with transistors, plenty of room for multiple orders of magnitude of optimization there. This is only a fraction of the future of physics-optimized compute.
Using size-1 piksters makes you really aware of all the subtle noise that your hidden plaque is gives your mind (I noticed they cleared up plaque un-reachable by floss+waterpiks+electric toothbrushes.. the first step to alignment/a faithful computation is reducing unnecessary noise (you notice this easily on microdoses of weed/psychedelics)
It’s a pareto-efficient improvement to give all alignment researchers piksters to eliminate this source of noise (align the aligners first—reducing unnecessary noise is always the first step to alignment [and near-term tFUS is also a means to reduce noise]). I know that one of the alignment offices had a lot of “freebies” that anyone could use—so piksters should be one of the useable freebies.
Lucy Lai’s new PhD thesis (and YouTube explainer) is really really worth reading/watching: https://x.com/drlucylai/status/1848528524790923669 and is more broadly relevant to people than most other PhD theses [esp on the original subject of making rational decisions under constraints of working memory].
What are some strategies you use to “reduce the hit” when you’re about to take in potentially bad news? This is important b/c it’s sometimes important to face up “bad news” earlier rather than later, and there is social loss in some people not being able to face it until it’s too late, esp b/c some kinds of “bad news” aren’t as incorrigible as they may initially appear (just that you need out-of-distribution strategies to make the proper amends)
[some examples of bad news: irreversible data loss, cancer diagnosis, elevated epigenetic age, loss of important friend, someone overpromised and underdelivered on you and that affects many of the promises you made]
[as AGI timelines come “nearer”, “bad news” may come at faster frequencies, but OOD ways to solve them may also come faster]
[sometimes you can ask yourself “how much wealth would you need to take in any bad news”]. Wealth is not fully-completely interchangeable with youth/intelligence/universal social acceptance, but it DEFINITELY has potential for tipping the needle..
Generally interesting people I wish more people would appreciate
Arthur Jiuliani (see his medium)/Adam Safron/Yohan John
Michael Woodley (kind of)
Jeremy Hadfield (https://imaginaries.substack.com/?utm_source=substack&utm_medium=web&utm_campaign=substack_profile)
Benjamin Anderson
http://augmentationlab.org (and their discord)
some neurofeedback peole (still TRYING THIS OUT, the S/N ratio might not be high, but finding the right person alone can make it super-worth it). Need to try LENS neurofeedback (money for this would be helpful btw)
Stephen Frey (undersells himself)
SOME quantified-self’ers
https://twitter.com/i/lists/1546283699829936129
MAYBES: https://publichealth.jhu.edu/faculty/2308/thomas-hartung (on brain organoids and autism), Bobby Azarian, https://www.informationphilosopher.com/solutions/scientists/layzer/
https://www.readcodon.com/p/machine
Interesting LessWrong people
dkirmani
matolcsid (infra-bayesianism I could never understand)
http://niplav.site/index.html (another gwern! calls himself midwit) Ad more interested in neurotech
Ege Erdil
https://gormful.net/ (gaspode, he has SO MUCH taste)
p.b.
https://www.lesswrong.com/users/bhauth
https://www.lesswrong.com/users/beren-1 (blog @ https://www.beren.io/ ). Janus likes him a lot
Matthew Barnett
Jacob Cannell
Interesting links:
https://www.lesswrong.com/posts/KQSpRoQBz7f6FcXt3
Could it be a good idea to enable a file uploading feature to LessWrong? (eg a file uploading feature of PDFs or certain images/media]. Could help against link rot, for example (and make posts from long ago last longer—I say this as someone who edits old posts to make them timeless).
I guess a problem here would be the legal issues around checking whether the file is copyright-protected.
Every single public mainstream AI model has RLHF’d out one of the most fundamental facts about human nature: that there exist vast differences between humans in basic ability/competence and they matter.
Is there a simple way to jailbreak the models, such as asking them to talk about a hypothetical parallel universe which is exactly like ours (same biology, same history), except that in the parallel universe humans can have different abilities and competences?
Are exotic computing paradigms (ECPs) pro-alignment?
cf https://twitter.com/niplav_site/status/1760277413907382685
They are orthogonal to the “scale is all you need” people, and the “scale is all you need” thesis is the hardest for alignment/interpretability
some examples of alternatives: https://www.lesswrong.com/posts/PyChB935jjtmL5fbo/time-and-energy-costs-to-erase-a-bit, Normal Computing, https://www.lesswrong.com/posts/ngqFnDjCtWqQcSHXZ/safety-of-self-assembled-neuromorphic-hardware, computing-related thiel fellows (eg Thomas Sohmers, Tapa Ghosh)
[this is also how to get into neglectedness again, which EA adopted as a principle but recently forgot]
from Charles Rosenbauer:
This is neat, but this does little to nothing to optimize non-AI compute. Modern CPUs are insanely wasteful with transistors, plenty of room for multiple orders of magnitude of optimization there. This is only a fraction of the future of physics-optimized compute.
Using size-1 piksters makes you really aware of all the subtle noise that your hidden plaque is gives your mind (I noticed they cleared up plaque un-reachable by floss+waterpiks+electric toothbrushes.. the first step to alignment/a faithful computation is reducing unnecessary noise (you notice this easily on microdoses of weed/psychedelics)
It’s a pareto-efficient improvement to give all alignment researchers piksters to eliminate this source of noise (align the aligners first—reducing unnecessary noise is always the first step to alignment [and near-term tFUS is also a means to reduce noise]). I know that one of the alignment offices had a lot of “freebies” that anyone could use—so piksters should be one of the useable freebies.
It’s Minifest day! Does anyone have any particular takes for the day? [pointers to important discussions/generators of energy/afterparties?]
Lucy Lai’s new PhD thesis (and YouTube explainer) is really really worth reading/watching: https://x.com/drlucylai/status/1848528524790923669 and is more broadly relevant to people than most other PhD theses [esp on the original subject of making rational decisions under constraints of working memory].
What are your scores on the US Economic Experts Comparison (Interactive Matrix)?
https://www.kentclarkcenter.org/economist-comparison-interactive-matrix/
Random content I’m reading (could be important)
https://research.vu.nl/en/persons/natalia-goriounova/publications/
Natalia Goriounova – Research output — Vrije Universiteit Amsterdam
https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(22)00208-X?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS136466132200208X%3Fshowall%3Dtrue
Evolution of cortical neurons supporting human cognition: Trends in Cognitive Sciences
https://academic.oup.com/cercor/article/32/11/2343/6373557
Verbal and General IQ Associate with Supragranular Layer Thickness and Cell Properties of the Left Temporal Cortex | Cerebral Cortex | Oxford Academic
https://www.frontiersin.org/articles/10.3389/fnhum.2019.00044/full
Frontiers | Genes, Cells and Brain Areas of Intelligence
https://academic.oup.com/cercor/article/25/12/4839/311644?login=false
Dendritic and Axonal Architecture of Individual Pyramidal Neurons across Layers of Adult Human Neocortex | Cerebral Cortex | Oxford Academic
https://academic.oup.com/cercor/article/33/6/2857/6633911?login=false
Strong and reliable synaptic communication between pyramidal neurons in adult human cerebral cortex | Cerebral Cortex | Oxford Academic
https://www.nature.com/articles/s41467-023-39946-9
Genes associated with cognitive ability and HAR show overlapping expression patterns in human cortical neuron types | Nature Communications
https://www.nature.com/articles/s41586-021-03813-8
Human neocortical expansion involves glutamatergic neuron diversification | Nature
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10345092/
Genes associated with cognitive ability and HAR show overlapping expression patterns in human cortical neuron types—PMC
https://www.esi-frankfurt.de/people/hermanncuntz/
Dr. Hermann Cuntz | Ernst Strüngmann Institute (ESI) for Neuroscience
https://www.cell.com/neuron/fulltext/S0896-6273(21)00625-5?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS0896627321006255%3Fshowall%3Dtrue
A general principle of dendritic constancy: A neuron’s size- and shape-invariant excitability: Neuron
https://www.pnas.org/doi/full/10.1073/pnas.1200430109
A scaling law derived from optimal dendritic wiring | PNAS
https://www.uni-giessen.de/de/fbz/zentren/icar3r/3r-symposium/speaker/name-2
Dr. Hermann Cuntz — 3R Symposium
https://scholar.google.com/citations?user=9g7Uj-MAAAAJ&hl=en
Hermann Cuntz - Google Scholar
https://www.biorxiv.org/content/10.1101/2023.02.27.530331v1
Topology recapitulates ontogeny of dendritic arbors | bioRxiv
https://www.treestoolbox.org/hermann/hermann_publications.html
Hermann Cuntz—homepage
https://www.biorxiv.org/content/10.1101/2023.03.15.532740v1.full
Skewed distribution of spines is independent of presynaptic transmitter release and synaptic plasticity and emerges early during adult neurogenesis | bioRxiv
https://www.treestoolbox.org/CNS2023_pareto_workshop/speakers.html
Optimality, evolutionary trade-offs, Pareto theory and degeneracy in neuronal modeling
https://www.biorxiv.org/content/10.1101/787911v1.full
A general principle of dendritic constancy – a neuron’s size and shape invariant excitability | bioRxiv
https://academic.oup.com/cercor/article/31/2/1008/5930850
Excess Neuronal Branching Allows for Local Innervation of Specific Dendritic Compartments in Mature Cortex | Cerebral Cortex | Oxford Academic
maybes
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4856390/
Optimal Current Transfer in Dendrites—PMC
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5474458/
Pyramidal Neurons in Different Cortical Layers Exhibit Distinct Dynamics and Plasticity of Apical Dendritic Spines—PMC
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6672209/
Dendritic Spikes in Apical Dendrites of Neocortical Layer 2⁄3 Pyramidal Neurons—PMC
https://www.nature.com/articles/s41598-017-09184-3
Branching morphology determines signal propagation dynamics in neurons | Scientific Reports
https://www.nature.com/articles/s41467-021-22741-9
Diversity amongst human cortical pyramidal neurons revealed via their sag currents and frequency preferences | Nature Communications
https://elifesciences.org/articles/46876
Cell-type specific innervation of cortical pyramidal cells at their apical dendrites | eLife
https://www.science.org/doi/full/10.1126/science.aax6239
Dendritic action potentials and computation in human layer 2⁄3 cortical neurons | Science
https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1011212
Biological complexity facilitates tuning of the neuronal parameter space | PLOS Computational Biology
https://alleninstitute.org/events/neuropixels-and-openscope-workshop/
2023 Neuropixels and OpenScope Workshop—Allen Institute
https://www.lifespan.io/news/extracellular-vesicles-from-stem-cells-reverse-senescence/
Extracellular Vesicles from Stem Cells Reverse Senescence
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8078853/
Classification of electrophysiological and morphological types in mouse visual cortex—PMC
https://www.biorxiv.org/content/10.1101/2020.03.31.018820v1.full
Human cortical expansion involves diversification and specialization of supragranular intratelencephalic-projecting neurons | bioRxiv
https://github.com/mousepixels/sanbomics_scripts/tree/main/RNAseq_method_comparison
sanbomics_scripts/RNAseq_method_comparison at main · mousepixels/sanbomics_scripts · GitHub
https://community.brain-map.org/t/introducing-the-allen-brain-cell-atlas/2444
Introducing the Allen Brain Cell Atlas! - How To / Allen Brain Cell (ABC) Atlas—Allen Brain Map Community Forum
https://knowledge.brain-map.org/data/LVDBJAW8BI5YSS1QUBG/summary
ABC Atlas—Mouse Whole Brain
https://community.brain-map.org/t/abc-atlas-user-guide-tools/2446
ABC Atlas User Guide: Tools—How To / Allen Brain Cell (ABC) Atlas—Allen Brain Map Community Forum
https://celltypes.brain-map.org/
Overview :: Allen Brain Atlas: Cell Types
https://portal.brain-map.org/explore/connectivity/synaptic-physiology/synaptic-physiology-experiment-methods/experimental-stimuli#intrinsic_stim
Synaptic Physiology Methods: Experimental Stimuli—brain-map.org
https://portal.brain-map.org/explore/connectivity/synaptic-physiology/synaptic-physiology-experiment-methods/cell-classification
Synaptic Physiology Methods: Cell Classification—brain-map.org
https://portal.brain-map.org/explore/connectivity/synaptic-physiology/synaptic-physiology-analysis-methods/synapse-characterization
Synaptic Physiology Methods: Synapse Characterization—brain-map.org
https://aisynphys.readthedocs.io/en/latest/matrix_analyzer.html#matrix-analyzer
Matrix Analyzer — aisynphys documentation
https://aisynphys.readthedocs.io/en/latest/matrix_analyzer.html#appendix
Matrix Analyzer — aisynphys documentation
https://portal.brain-map.org/explore/connectivity/synaptic-physiology/synaptic-physiology-analysis-methods/synapse-characterization#stp
Synaptic Physiology Methods: Synapse Characterization—brain-map.org
http://casestudies.brain-map.org/ggb#section_explorea
The Genetic Geography of the Brain :: Allen Institute for Brain Science
https://knowledge.brain-map.org/data?filter.program.title=CONTAINS~Allen%20Brain%20Map&limit=25&offset=0&sort=species.name~ASC
Data Catalog for Allen Institute’s Brain Knowledge Platform
https://celltypes.brain-map.org/
Overview :: Allen Brain Atlas: Cell Types
https://community.brain-map.org/t/how-to-download-raw-data-from-neuropixels-public-datasets/1923
How to download raw data from Neuropixels public datasets—How To / Brain Knowledge Platform—Allen Brain Map Community Forum
https://community.brain-map.org/t/new-signaling-mechanism-in-catecholaminergic-neurons/2507
New signaling mechanism in catecholaminergic neurons—OpenScope—Allen Brain Map Community Forum
https://www.science.org/doi/full/10.1126/science.aax6239
Dendritic action potentials and computation in human layer 2⁄3 cortical neurons | Science
https://www.sciencedirect.com/science/article/pii/S2211124719305777#bib48
Dissecting Sholl Analysis into Its Functional Components—ScienceDirect
https://scholar.google.com/citations?view_op=view_citation&hl=en&user=bcV-p4MAAAAJ&citation_for_view=bcV-p4MAAAAJ:YsMSGLbcyi4C
View article
Polls/surveys thread
Who do you subscribe to on reddit?
What are your scores on humanbenchmark.com?