text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# Instrumental convergence is what makes general intelligence possible
*TL;DR: General intelligence is possible because solving real-world problems requires solving common subtasks. Common subtasks are what give us instrumental convergence. Common subtasks are also what make AI useful; you* want *AIs to pursue instrum... | https://www.lesswrong.com/posts/GZgLa5Xc4HjwketWe/instrumental-convergence-is-what-makes-general-intelligence |
# Speculation on Current Opportunities for Unusually High Impact in Global Health
*Epistemic Status: armchair speculation from a non-expert.*
Short version: I expect things to get pretty bad in the Sahel region over the next year in particular. The area is an obvious target for global health interventions even in goo... | https://www.lesswrong.com/posts/rjjwfTgxkPJCyTasZ/speculation-on-current-opportunities-for-unusually-high |
# Internalizing the damage of bad-acting partners creates incentives for due diligence
Suppose you are cooperating with someone. It seems like there would be good reason to keep an eye on your partner to make sure that your partner does not do very bad things. For example, here are some reasons to keep an eye on your ... | https://www.lesswrong.com/posts/dXXzTAeBNKdXah3gh/internalizing-the-damage-of-bad-acting-partners-creates |
# "Rudeness", a useful coordination mechanic
I think the concept of "rudeness" is underappreciated. (Or, if people are appreciating it, they're doing so quietly where I can't find out about it)
I think a lot of coordination-social-tech relies on there being some kind of social karmic balance. A lot of actions aren't ... | https://www.lesswrong.com/posts/kJBKJJsurYiHadfQp/rudeness-a-useful-coordination-mechanic |
# Vanessa Kosoy's PreDCA, distilled
*This is an expanded version of my answer to* [*application problem 3*](https://docs.google.com/document/d/1NVVtdsfz7HiseVFSk3jYly4sPG4dG03wFFDrD8rBXU0/edit) *for Nate Soares and Vivek Hebbar's* [*SERI MATS*](https://www.lesswrong.com/posts/iR4kGzrWEJpXJ39ZB/seri-mats-program-winter... | https://www.lesswrong.com/posts/EFrJdhKPZXa4MA3Gr/vanessa-kosoy-s-predca-distilled |
# User-Controlled Algorithmic Feeds
I've been [playing with](https://www.jefftk.com/p/mastodon-linking-norms) [Mastodon more](https://www.jefftk.com/p/trying-mastodon), and the main way you see what other people have posted is by reading your feed. This consists of everything any of the people you follow has posted or... | https://www.lesswrong.com/posts/CqDDT8B2XBYfyBaEK/user-controlled-algorithmic-feeds |
# ACX Montreal Meetup November 26 2022
**Venue: **
McLennan Library Building at McGill. [3459 Mc Tavish St](https://www.google.com/maps/place/3459+Mc+Tavish+St,+Montreal,+QC+H3A+0C9/@45.5034767,-73.5791178,17z/data=!3m1!4b1!4m5!3m4!1s0x4cc91a47235d8f31:0xad33a8e94ffb9c43!8m2!3d45.503473!4d-73.5769291)
Room: HSSL - ... | https://www.lesswrong.com/events/7Gtb3qn3bcrwPk4KP/acx-montreal-meetup-november-26-2022 |
# Ways to buy time
[In our last post](https://www.lesswrong.com/posts/BbM47qBPzdSRruY4z/instead-of-technical-research-more-people-should-focus-on), we claimed:
> **On the margin, we think more alignment researchers should work on “buying time” interventions instead of technical alignment research **(or whatever else ... | https://www.lesswrong.com/posts/bkpZHXMJx3dG5waA7/ways-to-buy-time |
# Musings on the appropriate targets for standards
Consider two norms for how to critique those who violate your standards:
L: Focusing critique on those in power, who have the greatest opportunity to cause harm.
R: Focusing critique on those who it most accurately applies to, who most often cause harm.
I think bo... | https://www.lesswrong.com/posts/a2wJkueonTqMvcQDa/musings-on-the-appropriate-targets-for-standards |
# Noting an unsubstantiated belief about the FTX disaster
There is a narrative about the FTX collapse that I have noticed emerging[^MISSING-ID] as a commonly-held belief, despite little concrete evidence for or against it. The belief goes something like this:
* Sam Bankman Fried did what he did primarily for the s... | https://www.lesswrong.com/posts/HMaBPzrnvg2WCw6KL/noting-an-unsubstantiated-belief-about-the-ftx-disaster |
# Characterizing Intrinsic Compositionality in Transformers with Tree Projections
From [Twitter](https://twitter.com/topofmlsafety/status/1591098437830725632)
> Transformers become more ‘tree-like’ over the course of training, representing their inputs in a more hierarchical way. The authors find this by projecting t... | https://www.lesswrong.com/posts/6rn2ToNDwue527KsH/characterizing-intrinsic-compositionality-in-transformers |
# What's the Alternative to Independence?
When I [talk about](https://www.jefftk.com/p/spectrum-of-independence) [teaching my kids](https://www.jefftk.com/p/perverse-independence-incentives) [to be independent](https://www.jefftk.com/p/growing-independence), trying to get them to where they can do more on their own, o... | https://www.lesswrong.com/posts/X2pQ7bcoCc7gYPN7a/what-s-the-alternative-to-independence |
# A short critique of Vanessa Kosoy's PreDCA
*This critique is an addendum to* [*my distillation*](https://www.lesswrong.com/posts/EFrJdhKPZXa4MA3Gr/vanessa-kosoy-s-predca-distilled?_ga=2.165436411.531524530.1668251277-1637474216.1662379051) *for* [*application problem 3*](https://docs.google.com/document/d/1NVVtdsfz7... | https://www.lesswrong.com/posts/FhKkFcojhKZt7nHzG/a-short-critique-of-vanessa-kosoy-s-predca-1 |
# The Futility of Status and Signalling
### I
In [The Futility of Emergence](https://www.lesswrong.com/posts/8QzZKw9WHRxjR4948/the-futility-of-emergence) Eliezer names a current theory which is as flawed as the theories of flogiston and vitalism. Theories that give mysterious answers which just appears to explain the... | https://www.lesswrong.com/posts/vJbczBHtDPmbdgCcY/the-futility-of-status-and-signalling |
# Announcing Nonlinear Emergency Funding
*\[Applications are now closed. Please* [*apply to Open Phil*](https://forum.effectivealtruism.org/posts/HPdWWetJbv4z8eJEe/open-phil-is-seeking-applications-from-grantees-impacted-by) *or* [*SFF*](https://forum.effectivealtruism.org/posts/buEazpmcKJhM5KRGx/sff-speculation-grant... | https://www.lesswrong.com/posts/SkoxdYCAozBPfcJrP/announcing-nonlinear-emergency-funding |
# In Defence of Temporal Discounting in Longtermist Ethics
Introduction
============
My thoughts on temporal discounting in longtermist ethics.
Two Senses of Normative Ethics
==============================
There are two basic senses of a normative moral system:
1\. Criterion of judgment: "what is right/good; wron... | https://www.lesswrong.com/posts/zFSnbXojsNS4725v2/in-defence-of-temporal-discounting-in-longtermist-ethics |
# Estimating the probability that FTX Future Fund grant money gets clawed back
It would be great to have more people involved in estimating the probability that FTX Future Fund grants end up getting clawed back (i.e., that projects that received such grants are asked to give the money back). This is a very important q... | https://www.lesswrong.com/posts/9weamhzzBpqxz3A2e/estimating-the-probability-that-ftx-future-fund-grant-money |
# AI Safety Microgrant Round
We are pleased to announce an AI Safety Microgrants Round, which will provide micro-grants to field-building projects and other initiatives that can be done with less.
We believe there are projects and individuals in the AI Safety space who lack funding but have high agency and potential... | https://www.lesswrong.com/posts/LCypC3mdEPxXmHLGg/ai-safety-microgrant-initiative |
# Why don't we have self driving cars yet?
1) Is it because of regulations?
2) Is it because robustness in the real world (or just robustness in general) turns out to be very hard for current AI systems, and robustness is much more important for self driving cars than other areas where we have seen more rapid AI prog... | https://www.lesswrong.com/posts/Mtmq827AE7PCdHzef/why-don-t-we-have-self-driving-cars-yet |
# Improving Emergency Vehicle Utilization
Every ambulance that goes out represents a massive waste of potential: it doesn't have any passengers! Ambulances need space to transport people on the return trip, and it really says something about humanity that we've gone so many years with that capacity completely idle on ... | https://www.lesswrong.com/posts/zQ3jhyAPoHiChdFjj/improving-emergency-vehicle-utilization |
# Two New Newcomb Variants
Two Newcomb variants to add to the list of examples where optimal choice and optimal policy are diammetrically opposed. I don't think problems these exist anywhere else yet.
### 4 Boxes Problem
In a game show there are 4 transparent boxes in a row, each of which starts off with $1 inside. ... | https://www.lesswrong.com/posts/Lhbkc8842L3GDDvtq/two-new-newcomb-variants |
# Will we run out of ML data? Evidence from projecting dataset size trends
**Summary**: Based on our previous analysis of trends in dataset size, we project the growth of dataset size in the language and vision domains. We explore the limits of this trend by estimating the total *stock* of available unlabeled data ove... | https://www.lesswrong.com/posts/Couhhp4pPHbbhJ2Mg/will-we-run-out-of-ml-data-evidence-from-projecting-dataset |
# Moral contagion heuristic
> "Remember that time you went roller-skating with pink eyebrows? I was embarrassed even though it wasn't me, but because you're my mom." - My 7 year old
A prevalent and arguably universal human cognitive bias is the contagion heuristic - the idea that mere contact or association with a ba... | https://www.lesswrong.com/posts/zjxJ9Fkpv9ZMdFxAi/moral-contagion-heuristic |
# Preventing, reversing, and addressing data leakage: some thoughts
In the last few months, I've been thinking about the problem of
accidental leakage of data and how to prevent it, reverse it, and
address its aftermath. This post includes various thoughts, some of
them including tool-specific guidance, and some of th... | https://www.lesswrong.com/posts/QTfwhsQkzGbKtAXZq/preventing-reversing-and-addressing-data-leakage-some |
# The economy as an analogy for advanced AI systems
Eric Drexler’s Comprehensive AI Services (CAIS), particularly as set out in his 2019 report [*Reframing Superintelligence*](https://www.fhi.ox.ac.uk/wp-content/uploads/Reframing_Superintelligence_FHI-TR-2019-1.1-1.pdf), is a complex model with many different assumpti... | https://www.lesswrong.com/posts/oH3XmScSFnZt6x2eN/the-economy-as-an-analogy-for-advanced-ai-systems-2 |
# Better Mastodon Aliases
When I [joined Mastodon](https://www.jefftk.com/p/trying-mastodon) I was `@jefftk@schelling.pt`. A few days ago that server had an incident and I moved to `[mastodon.mit.edu](https://mastodon.mit.edu)`, becoming `@jefftk@mastodon.mit.edu`. This got me thinking about a wart in the current Mast... | https://www.lesswrong.com/posts/at8QigBZnxirgHraT/better-mastodon-aliases |
# Internal communication framework
Usually when we think about humans and agency, we focus on the layer where one human is one agent.
, how do we make sense of it? How do we sort out all the disparate urges, emotions, and preferences, and compress them into legibl... | https://www.lesswrong.com/posts/kmpNkeqEGvFue7AvA/value-formation-an-overarching-model |
# Entropy Scaling And Intrinsic Memory
This is the second post in a series that explores a circle of ideas around entropy and information processing in complex systems. We hope it will form an accessible introduction to [James Crutchfield's Computational Mechanics](http://csc.ucdavis.edu/~cmg/compmech/pubs/CalcEmerg... | https://www.lesswrong.com/posts/tcicoJatbBMpsDbSW/entropy-scaling-and-intrinsic-memory |
# Why bet Kelly?
Suppose you know what outcomes are better than what other outcomes, but not by how much. Given two possible actions you could take, $A$ and $B$, if you know what probability distributions over outcomes each of them results in, this doesn't necessarily help you pick which is better. Being able to compa... | https://www.lesswrong.com/posts/NPzGfDi3zMJfM2SYe/why-bet-kelly |
# The limited upside of interpretability
**TL;DR**: A strategy aiming to elicit latent knowledge (or to make any hopefully robust, hopefully generalizable prediction) from interpreting an AGI’s fine-grained internal data may be unlikely to succeed, given that the complex system of an AGI’s agent-environment interactio... | https://www.lesswrong.com/posts/bkjoHFKjRJhYMebXr/the-limited-upside-of-interpretability |
# Some research ideas in forecasting
In the past, I have researched how we can effectively [pool the predictions of many experts](https://forum.effectivealtruism.org/s/hjiBqAJNKhfJFq7kf). For the most part, I am now focusing on directing [Epoch](https://epochai.org/) and AI forecasting.
However, I have accumulated a... | https://www.lesswrong.com/posts/AbDpemQiNFw8cXyEy/some-research-ideas-in-forecasting |
# SBF x LoL
_Epistemic status: community conversation starter._
_Content warnings: mental illness, IQ, drug abuse, speculation about unproven events, the concept of permanent mental degradation, flashing lights / loud sounds on a linked video, a link to the LessWrong tag about (discussion of the meta-idea of) infohaz... | https://www.lesswrong.com/posts/64bX5saWkHY4tMDRS/sbf-x-lol |
# What is our current best infohazard policy for AGI (safety) research?
I struggle to see what the infohazard policy for AGI safety research writing should be. For example, it has been pointed out multiple times that GPT-3 paper was an infohazard, because so many people realised this kind of model was possible, and de... | https://www.lesswrong.com/posts/eojjuQEpas6LRDmj6/what-is-our-current-best-infohazard-policy-for-agi-safety |
# Is the speed of training large models going to increase significantly in the near future due to Cerebras Andromeda?
Cerebras recently unveiled Andromeda - [https://www.cerebras.net/andromeda/,](https://www.cerebras.net/andromeda/,) an AI supercomputer that enables near linear scaling. Do I understand correctly that ... | https://www.lesswrong.com/posts/vuRNYiekLSABSJPGg/is-the-speed-of-training-large-models-going-to-increase |
# The FTX Saga - Simplified
*Disclaimer: This is an evolving situation. It might take months to get an accurate picture of what exactly happened. This post is my attempt to simplify what is currently the general theory of what happened at FTX as of November 16, 2022. The goal of this post is to explain to those not we... | https://www.lesswrong.com/posts/cM4KWHoeWdgeNuQa8/the-ftx-saga-simplified |
# EA & LW Forums Weekly Summary (7th Nov - 13th Nov 22')
*Supported by Rethink Priorities*
This is part of a weekly series summarizing the top (40+ karma) posts on the EA and LW forums - you can see the full collection [here.](https://forum.effectivealtruism.org/s/W4fhpuN26naxGCBbN) The first post includes some detai... | https://www.lesswrong.com/posts/RwkjcjqfnZmEJCK54/ea-and-lw-forums-weekly-summary-7th-nov-13th-nov-22 |
# Developer experience for the motivation
I watched this [YouTube video](https://www.youtube.com/watch?v=beP0QXH4Ta0) a few months ago that really stuck with me. It was about how to organize your kitchen. The video went through 10 tips, but before getting in to those tips it talked about the why. Why is organizing you... | https://www.lesswrong.com/posts/xzJHpLii7g2exFmDP/developer-experience-for-the-motivation |
# The two conceptions of Active Inference: an intelligence architecture and a theory of agency
I think much of the confusion about Active Inference arises because the term is used to refer to two related, yet distinct concepts (one can also call them abstractions, theories, or ontics): an *intelligence architecture* a... | https://www.lesswrong.com/posts/2BPPwboTDrAMFiGHe/the-two-conceptions-of-active-inference-an-intelligence |
# Unpacking "Shard Theory" as Hunch, Question, Theory, and Insight
I read several [shard theory posts](https://www.lesswrong.com/s/nyEFg3AuJpdAozmoX) and found the details interesting, but I couldn't quite see the big picture. I'm used to hearing "theory" refer to [*a falsifiable generalization of data*](https://en.wi... | https://www.lesswrong.com/posts/BXzhtfXeP8WwdQTXy/unpacking-shard-theory-as-hunch-question-theory-and-insight |
# Current themes in mechanistic interpretability research
*This post gives an overview of discussions - from the perspective and understanding of the interpretability team at Conjecture - between mechanistic interpretability researchers from various organizations including Conjecture, Anthropic, Redwood Research, Open... | https://www.lesswrong.com/posts/Jgs7LQwmvErxR9BCC/current-themes-in-mechanistic-interpretability-research |
# Disagreement with bio anchors that lead to shorter timelines
*This would have been a submission to the *[*FTX AI worldview prize*](https://ftxfuturefund.org/announcing-the-future-funds-ai-worldview-prize/)*. I’d like to thank Daniel Kokotajlo, Ege Erdil, Tamay Besiroglu, Jaime Sevilla, Anson Ho, Keith Wynroe, Pablo ... | https://www.lesswrong.com/posts/Q3fesop6HKnemJ5Jc/disagreement-with-bio-anchors-that-lead-to-shorter-timelines |
# If Professional Investors Missed This...
One of the largest cryptocurrency exchanges, [FTX](https://en.wikipedia.org/wiki/FTX_(company)), recently imploded after [apparently](https://www.nytimes.com/2022/11/14/technology/ftx-sam-bankman-fried-crypto-bankruptcy.html?unlocked_article_code=40IPATU-4dJKAPYsqyYO1YAmhYMmQ... | https://www.lesswrong.com/posts/gLdZEzC3afQn2Fon5/if-professional-investors-missed-this |
# Questions about Value Lock-in, Paternalism, and Empowerment
Pulling a child out of the path of a fast car is the right thing to do, whether or not the child agrees, understands, is grateful, or even is hurt during the rescue. [Paternalistic](https://plato.stanford.edu/archives/sum2014/entries/paternalism/) acts like... | https://www.lesswrong.com/posts/nLjtqdhRaKcEGb4NA/questions-about-value-lock-in-paternalism-and-empowerment |
# When should we be surprised that an invention took “so long”?
My first highly popular essay was “[Why did we wait so long for the bicycle?](https://rootsofprogress.org/why-did-we-wait-so-long-for-the-bicycle)” I’ve asked the same question of the [cotton gin](https://rootsofprogress.org/why-did-we-wait-so-long-for-th... | https://www.lesswrong.com/posts/fJvuhwGiunqwfDAwb/when-should-we-be-surprised-that-an-invention-took-so-long |
# Is there some reason LLMs haven't seen broader use?
When GPT-3 first came out, I expected that people would use it as a sort of "common-sense reasoning module". That is, if you want to process or generate information in some way, then you can give GPT-3 a relevant prompt, and repeatedly apply it to a bunch of differ... | https://www.lesswrong.com/posts/nZBfpwRnZ63xF67hQ/is-there-some-reason-llms-haven-t-seen-broader-use |
# Kelsey Piper's recent interview of SBF
**Edit:** See the bottom for some important updates.
Kelsey Piper from Vox's Future Perfect very recently released an interview (made through Twitter DMs) with Sam Bankman-Fried. The interview goes in depth into the events surrounding FTX and Alameda Research.
> As we message... | https://www.lesswrong.com/posts/Hyk7ier98DPfZfzcf/kelsey-piper-s-recent-interview-of-sbf |
# Why are profitable companies laying off staff?
Consider all the reasons you might have to reduce headcount in an economic downturn:
1. You rely on venture capital to survive which is no longer as easy to get, so you simply can't afford to pay your staff.
2. You've started making a loss, so you need to fire people t... | https://www.lesswrong.com/posts/sqModqWGJRn3YJghi/why-are-profitable-companies-laying-off-staff |
# The Ground Truth Problem (Or, Why Evaluating Interpretability Methods Is Hard)
Work done @ [SERI-MATS](https://www.serimats.org/).
Evaluating interpretability methods (and so, developing good ones) is really hard because we have no ground truth. Or at least, no ground truth that we can compare our interpretations d... | https://www.lesswrong.com/posts/snbNNQSG35D5XHtpn/the-ground-truth-problem-or-why-evaluating-interpretability |
# Deontology and virtue ethics as "effective theories" of consequentialist ethics
This is a rough attempt to present deontology and virtue ethics as approximate, effective theories [^7mf0kdky41t] of consequentialist morality, for agents severely bounded in their ability to determine the consequences of their actions,... | https://www.lesswrong.com/posts/JvXNCc6w58iHTYHuz/deontology-and-virtue-ethics-as-effective-theories-of |
# Sadly, FTX
It has been quite a past two weeks, with different spheres in [deeply divided narratives](https://twitter.com/WilliamAEden/status/1591334979749154816). In addition to the liberation of Kherson City, the midterm elections that _of course_ took forever to resolve and the ongoing hijinks of Elon Musk taking ... | https://www.lesswrong.com/posts/p4XpZWcQksSiCPG72/sadly-ftx |
# Covid 11/17/22: Slow Recovery
While the news has been fast and furious on many fronts, Covid is not one of them.
Except that, this week, I got it.
On Monday evening, I started having chills. They seem better during the day, worse at night, slowly improving.
That night, I was almost entirely unable to sleep. I wou... | https://www.lesswrong.com/posts/GGfgtKKTAda6NDPCw/covid-11-17-22-slow-recovery |
# Results from the interpretability hackathon
We ran a [mechanistic interpretability](https://www.lesswrong.com/posts/Jgs7LQwmvErxR9BCC/current-themes-in-mechanistic-interpretability-research) [hackathon](https://itch.io/jam/interpretability) ([original post](https://forum.effectivealtruism.org/posts/vxLrFdrqRPdaHJwgs... | https://www.lesswrong.com/posts/hhhmcWkgLwPmBuhx7/results-from-the-interpretability-hackathon |
# AI Forecasting Research Ideas
**Overview**
* The linked document contains a collection of AI Forecasting research ideas, prepared by some [Epoch](https://epochai.org/) employees in a personal capacity
* We think that these are interesting and valuable projects that research interns or students could look into, ... | https://www.lesswrong.com/posts/ACcWrHRmtfgHdz46y/ai-forecasting-research-ideas |
# Announcing the Progress Forum
I’d like to invite you to join the [Progress Forum](https://progressforum.org/), the new online home for the progress community. It's a clone of this site, but with a focus on [progress studies](https://rootsofprogress.org/progress-studies-a-moral-imperative) and the [philosophy of prog... | https://www.lesswrong.com/posts/ubs5q4MpsgWBr2Tzw/announcing-the-progress-forum |
# Mastodon Replies as Comments
The comment section on most blogs is pretty minimal, with the real discussion happening elsewhere, but people who come to the post later won't see that discussion. One of the more unusual choices I've made with this blog is that instead of hosting comments here, I pull in and display com... | https://www.lesswrong.com/posts/sE5wHZCDyzCSuwBxd/mastodon-replies-as-comments |
# LLMs may capture key components of human agency
TLDR: LLM confabulation may be a significant feature - not flaw - of how human memory works and how humans experience agency in the world. Namely through narratives that are largely self-consistent - but not guaranteed to match facts in the world.
**Introduction**
Th... | https://www.lesswrong.com/posts/ZXB3HbAuwJakwjPB6/llms-may-capture-key-components-of-human-agency |
# Engineering Monosemanticity in Toy Models
Overview
========
In some neural networks, individual neurons correspond to natural "features" in the input. Such *monosemantic* neurons are much easier to interpret, because in a sense they only do one thing. By contrast, some neurons are *polysemantic*, meaning that they... | https://www.lesswrong.com/posts/LvznjZuygoeoTpSE6/engineering-monosemanticity-in-toy-models |
# Don't design agents which exploit adversarial inputs
**Summary.** Consider two common alignment design patterns:
1. Optimizing for the output of a grader which evaluates plans, and
2. Fixing a utility function and then argmaxing over all possible plans.
These design patterns incentivize the agent to find adver... | https://www.lesswrong.com/posts/jFCK9JRLwkoJX4aJA/don-t-design-agents-which-exploit-adversarial-inputs |
# Halifax, NS – Monthly Rationalist, EA, and ACX Meetup
In the interest of growing the LW/EA/ACX community in Halifax, we are hosting meetups on the last Saturday of each month. The format is friendly, unstructured discussion (future iterations could emphasise particular topics). Old hands and newcomers are welcome!
... | https://www.lesswrong.com/events/zrWjyuRMWAqW3YxNj/halifax-ns-monthly-rationalist-ea-and-acx-meetup |
# Updates on scaling laws for foundation models from ' Transcending Scaling Laws with 0.1% Extra Compute'
I am not sure if this paper is flying under the radar for many people, but has anyone read [Transcending Scaling Laws with 0.1% Extra Compute](https://arxiv.org/pdf/2210.11399.pdf)? If so, how do you think it comp... | https://www.lesswrong.com/posts/RDvuJaamWtT4JbsSr/updates-on-scaling-laws-for-foundation-models-from |
# Contra Chords
When you see contra dance tunes or other similar music written down there are usually chords along with the tune, but I rarely like them. Some of this is aesthetic differences, where the chords are often written in a style where you switch chords rapidly to closely follow the melody, but the real probl... | https://www.lesswrong.com/posts/igkbHD99s5md9m3XC/contra-chords |
# Distillation of "How Likely Is Deceptive Alignment?"
Summary
=======
Deceptive alignment seems to be the likely outcome of training a sufficiently intelligent AI using gradient descent. An ML model with a long term goal and an understanding of its training process would probably pretend to be aligned during traini... | https://www.lesswrong.com/posts/XKraEJrQRfzbCtzKN/distillation-of-how-likely-is-deceptive-alignment |
# Is there any policy for a fair treatment of AIs whose friendliness is in doubt?
A game theory question.
If there ever exists a roughly-human-level agenty AI who could grow to overpower humans but who humans have an opportunity to stop because takeoff is slow enough. Assume the AI could coexist with humanity but fea... | https://www.lesswrong.com/posts/y5k77ZEyrAzY8E8EK/is-there-any-policy-for-a-fair-treatment-of-ais-whose |
# The Disastrously Confident And Inaccurate AI
*The following is my first post on AI Safety, in a manner of speaking. To be more precise, this is my distillation of the first post in the Risks from Learned Optimization sequence by Evan Hubringer et al. Further distillations will follow. (I do not mean to sound threat... | https://www.lesswrong.com/posts/WNTGe87fHwDZMLqzW/the-disastrously-confident-and-inaccurate-ai |
# Reflective Consequentialism
*Epistemic Status: My position makes sense to me and seems pretty plausible. However, I haven't thought too hard about it, nor have I put it to the test by discussing with others. Putting it to the test is what I'm trying to do here. I also don't have the best understanding of what virtue... | https://www.lesswrong.com/posts/xYYrADpKD22bH6F6o/reflective-consequentialism |
# Choosing the right dish
1
=
From what I understand, despite its simplicity, there aren't many dishes held in higher esteem by the culinary community[^vrxush0qt1] than the French omelette.
To all of you uncultured numskulls out there who don't spend hours upon hours watching YouTube videos about cooking like I do, ... | https://www.lesswrong.com/posts/jLXxCW2eiXyc7TQXe/choosing-the-right-dish |
# By Default, GPTs Think In Plain Sight
*Epistemic status: Speculation with some factual claims in areas I’m not an expert in.*
*Thanks to Jean-Stanislas Denain, Charbel-Raphael Segerie, Alexandre Variengien, and Arun Jose for helpful feedback on drafts, and thanks to janus, who shared related ideas.*
Main claims
==... | https://www.lesswrong.com/posts/bwyKCQD7PFWKhELMr/by-default-gpts-think-in-plain-sight |
# A Short Dialogue on the Meaning of Reward Functions
*Produced as part of the* [*SERI ML Alignment Theory Scholars Program*](https://serimats.org/) *\- Winter 2022 Cohort.*
The following is a short slack dialogue between Leon Lang, Quintin Pope, and Peli Grietzer that emerged as part of the [SERI-MATS](https://www.s... | https://www.lesswrong.com/posts/Hx48HgHzDTsSFoJui/a-short-dialogue-on-the-meaning-of-reward-functions |
# Elastic Productivity Tools
Like most tech bros, I’m a little too interested in productivity and optimizing my life. I’ve even made a few of my own tools to help me stay focused and efficient. In the process of trying to find and build the best productivity tools possible, I discovered common elements among all the p... | https://www.lesswrong.com/posts/eJYDck4EthKFLga7q/elastic-productivity-tools |
# Make the Drought Evaporate!
~ *desalinate with <5% the energy & equipment* ~
**TL;DR —** We need to stop droughts, especially as climate change makes them worse. Desalination is nice, capturing clean water from the sea, yet it requires expensive equipment and lots and lots of energy. We can, instead, ***accelerate ... | https://www.lesswrong.com/posts/wttyWyKqkPyncuoCm/make-the-drought-evaporate |
# Update to Mysteries of mode collapse: text-davinci-002 not RLHF
I (and many others) did not realize this before, but: `text-davinci-002` and `text-davinci-001`, the InstructGPT models on the OpenAI API, were not trained with RLHF (reinforcement learning from human feedback) as described in the [InstructGPT paper](ht... | https://www.lesswrong.com/posts/mbGjzyy6eJXT4gFpm/update-to-mysteries-of-mode-collapse-text-davinci-002-not |
# ARC paper: Formalizing the presumption of independence
*(I did not have anything to do with this paper and these are just my own takes.)*
The Alignment Research Center recently published their second report, [Formalizing the presumption of independence](https://arxiv.org/abs/2211.06738). While it's not explicitly a... | https://www.lesswrong.com/posts/wG5KTj5jFiibydgmk/arc-paper-formalizing-the-presumption-of-independence |
# Decision Theory but also Ghosts
*Spoiler Warning:* [*The Sixth Sense (1999)*](https://www.imdb.com/title/tt0167404/) *is a good movie. Watch it before reading this.*
A much smaller eva once heard of Descartes' *Cogito ergo sum* as being the pinnacle of skepticism, and disagreed. *"Why couldn't I doubt that? Maybe I... | https://www.lesswrong.com/posts/wjA6vAnTWxJSQKadK/decision-theory-but-also-ghosts |
# Career Scouting: Dentistry
As a high school student, I worry a great deal over my future profession. According to Cal Newport, career satisfaction for any choice of occupation often won't materialize until you've become "so good they can't ignore you" at what you do. Based on this, Newport recommends directing your ... | https://www.lesswrong.com/posts/odh2co2fLLzSGC4Zx/career-scouting-dentistry |
# Limits to the Controllability of AGI
**Roman Yampolskiy**
**(Summary by Karl von Wendt and Remmelt Ellen)**
*This is a summary of the paper *[***“On the Controllability of Artificial Intelligence: An Analysis of Limitations”***](https://journals.riverpublishers.com/index.php/JCSANDM/article/view/16219/13165)*, ... | https://www.lesswrong.com/posts/Zq4SxjwKBB6Ld3SNf/limits-to-the-controllability-of-agi |
# Scott Aaronson on "Reform AI Alignment"
The framing is based on contrasting Orthodox Judaism with Reform Judaism, only for AI Alignment.
> If AI alignment is a religion, it’s now large and established enough to have a thriving “Reform” branch, in addition to the original “Orthodox” branch epitomized by Eliezer Yudk... | https://www.lesswrong.com/posts/Zqk4FFif93gvquAnY/scott-aaronson-on-reform-ai-alignment |
# Here's the exit.
There's a kind of game here on Less Wrong.
It's the kind of game that's a little rude to point out. Part of how it works is by not being named.
Or rather, attempts to name it get dissected so everyone can agree to continue ignoring the fact that it's a game.
So I'm going to do the rude thing. But... | https://www.lesswrong.com/posts/kcoqwHscvQTx4xgwa/here-s-the-exit |
# Interview with Matt Freeman
An interview with Matt Freeman, co-founder of the [Guild of the ROSE](https://guildoftherose.org/), an organization for rationalist community and education. Topics discussed include:
* The motivation and vision behind creating the Guild.
* What might be possible with consistent ratio... | https://www.lesswrong.com/posts/GuHWD8R2MQtFy4Hbh/interview-with-matt-freeman |
# Utilitarianism Meets Egalitarianism
This post is mostly propaganda for the Nash Bargaining solution, but also sets up some useful philosophical orientation. This post is also the first post in my [geometric rationality](https://www.lesswrong.com/s/4hmf7rdfuXDJkxhfg) sequence.
Utilitarianism
==============
Let's pr... | https://www.lesswrong.com/posts/hTMFt3h7QqA2qecn7/utilitarianism-meets-egalitarianism |
# [Hebbian Natural Abstractions] Introduction
With this sequence, we (Sam + Jan) want to provide a principled derivation of the [natural abstractions hypothesis](https://www.alignmentforum.org/posts/Nwgdq6kHke5LY692J/alignment-by-default#Unsupervised__Natural_Abstractions) (which we will introduce in-depth in later po... | https://www.lesswrong.com/posts/mFCbW6rYLzARqi5pf/hebbian-natural-abstractions-introduction |
# Miscellaneous First-Pass Alignment Thoughts
I have been learning more about alignment theory in the last couple of months, and have heard from many people that writing down [naive hypotheses](https://www.lesswrong.com/posts/ubdp8qAL8Gfki2pYo/naive-hypotheses-on-ai-alignment) can be a good strategy for developing you... | https://www.lesswrong.com/posts/iEbFJqQmC9ejxrcEN/miscellaneous-first-pass-alignment-thoughts |
# LessWrong readers are invited to apply to the Lurkshop
**Summary**: We’re running a free retreat for LessWrong users (lurkers) who aren’t yet working on existential risk mitigation but might be interested, in the hopes of catalyzing some of them into people who are. It’s called the Lurkshop. [**Apply here**](https:/... | https://www.lesswrong.com/posts/r8K8ddYDYNrPJRFyC/lesswrong-readers-are-invited-to-apply-to-the-lurkshop |
# Austin LW meetup notes: The FTX Affair
This article is a writeup of the conversation at a [meetup](https://austinlesswrong.com/2022/11/16/cafe-meetup-11-19-the-ftx-exchange-collapse/) hosted by [Austin Less Wrong](https://www.lesswrong.com/groups/syf4gEbaNQ8btKCbb) on Saturday, November 19, 2022. The topic was the c... | https://www.lesswrong.com/posts/tpK3nmk4fFDJArGdC/austin-lw-meetup-notes-the-ftx-affair |
# Meta AI announces Cicero: Human-Level Diplomacy play (with dialogue)
**Abstract**
Despite much progress in training AI systems to imitate human language, building agents that use language to communicate intentionally with humans in interactive environments remains a major challenge. We introduce Cicero, the first ... | https://www.lesswrong.com/posts/3TCYqur9YzuZ4qhtq/meta-ai-announces-cicero-human-level-diplomacy-play-with |
# Simple Improvement to College Football Overtime Rules
For a long time I have thought that a simple tweak to the NCAAF overtime rules would fix several problems with no downsides. The Alabama-LSU game reinforced my impression of this by showing what we could have every time.
**PROPOSED: In overtime, teams cannot kic... | https://www.lesswrong.com/posts/Him8wentkaMeMACqW/simple-improvement-to-college-football-overtime-rules |
# A Walkthrough of In-Context Learning and Induction Heads (w/ Charles Frye) Part 1 of 2
New paper walkthrough: [In-Context Learning and Induction Heads](https://www.youtube.com/watch?v=dCkQQYwPxdM). This is the second paper in [Anthropic's Transformer Circuits thread](https://transformer-circuits.pub/), a series of p... | https://www.lesswrong.com/posts/xhkxv6qnnGmqmxxsz/a-walkthrough-of-in-context-learning-and-induction-heads-w |
# Tyranny of the Epistemic Majority
This post is going to mostly be propaganda for Kelly betting. However, the reasons presented in this post differ greatly from the reasons people normally use to argue for Kelly betting.
The Steward of Myselves
=======================
The curse of uncertainty is that I must make de... | https://www.lesswrong.com/posts/o3RLHYviTE4zMb9T9/tyranny-of-the-epistemic-majority |
# Progress links and tweets, 2022-11-22
**The Progress Forum**
----------------------
* [Welcome to the Progress Forum, the online home for the progress community](https://progressforum.org/posts/XTJfsw35jJcrdSECT/welcome-to-the-progress-forum)
* [How fear of the ‘Mega-Machine’ helped end America’s postwar Golden... | https://www.lesswrong.com/posts/DcgC4ADsrmbGNvvjT/progress-links-and-tweets-2022-11-22 |
# AI will change the world, but won’t take it over by playing “3-dimensional chess”.
By [Boaz Barak](https://www.boazbarak.org/) and [Ben Edelman](https://www.benjaminedelman.com/)
*\[Cross-posted on* [*Windows on Theory blog*](https://windowsontheory.org/?p=8469)*; See also Boaz’s posts on *[*longtermism*](https://w... | https://www.lesswrong.com/posts/zB3ukZJqt3pQDw9jz/ai-will-change-the-world-but-won-t-take-it-over-by-playing-3 |
# Human-level Full-Press Diplomacy (some bare facts).
Key links?
----------
* Here's the paper: [https://www.science.org/doi/10.1126/science.ade9097](https://www.science.org/doi/10.1126/science.ade9097)
* Here's the blog post: [https://ai.facebook.com/blog/cicero-ai-negotiates-persuades-and-cooperates-with-people... | https://www.lesswrong.com/posts/oktnxsng7Dbc4aoZP/human-level-full-press-diplomacy-some-bare-facts |
# Announcing AI Alignment Awards: $100k research contests about goal misgeneralization & corrigibility
*We’re grateful to our advisors Nate Soares, John Wentworth, Richard Ngo, Lauro Langosco, and Amy Labenz. We're also grateful to Ajeya Cotra and Thomas Larsen for their feedback on the contests. *
**TLDR:** [**AI Al... | https://www.lesswrong.com/posts/JNtGxrusJRpx53Q8L/announcing-ai-alignment-awards-usd100k-research-contests |
# Feeling Old: Leaving your 20s in the 2020s
A younger friend recently asked why I (and many others my age) make so many comments about being old. After some thought, I realized a big part of this was related to experiencing specific meaningful life stages during the pandemic.
In 2019, I could (and often did) stay up... | https://www.lesswrong.com/posts/vBkPxyCuAjiXNDiFg/feeling-old-leaving-your-20s-in-the-2020s |
# Rationalist Town Hall: FTX Fallout Edition (RSVP Required)
*Edit: K, all those who filled out the form should have emails now. No promises that any further form-fills will get emails.*
*Stated at the top for emphasis: **you have to fill out*** [***the RSVP form***](https://docs.google.com/forms/d/e/1FAIpQLScEiU0Rdi... | https://www.lesswrong.com/posts/GhCiaess4ATKpjyk6/rationalist-town-hall-ftx-fallout-edition-rsvp-required |
# Simulators, constraints, and goal agnosticism: porbynotes vol. 1
*This is a part of a maybe-series where I braindump safety notes while waiting on training runs to complete. It's mostly talking to myself, but talking to myself in public seems somewhat more productive. The content of this post is not guaranteed to be... | https://www.lesswrong.com/posts/DSEwkvj8W7y8C3jau/simulators-constraints-and-goal-agnosticism-porbynotes-vol-1 |
# Sets of objectives for a multi-objective RL agent to optimize
**Background: A multi-objective decision-making AI**
----------------------------------------------------
Previously I've proposed[ balancing multiple objectives via multi-objective RL](https://www.lesswrong.com/posts/HoQ8WaEHXdkaMbpzx/can-we-achieve-agi... | https://www.lesswrong.com/posts/4mvdZXjwJHv9tSAWB/sets-of-objectives-for-a-multi-objective-rl-agent-to-1 |
# Human-level Diplomacy was my fire alarm
Epistemic status: AI is my hobby, not my job.
Meta just released their new Diplomacy AI, Cicero. Sure, it was several different modules kludged together in a somewhat clever way, with no fundamentally new ideas or massive scaling, but it was damn impressive.
What surprises ... | https://www.lesswrong.com/posts/AZHHEPYWvTovvtikz/human-level-diplomacy-was-my-fire-alarm |
# Announcing AI safety Mentors and Mentees
Executive summary
=================
* AI safety Mentors and Mentees is a program that matches mentors with mentees to accelerate their AI safety journey. Our aim is to get more people into full-time AI safety roles faster. We expect nearly all mentees to do this next to th... | https://www.lesswrong.com/posts/EjX63wQoMSoCHMrmY/announcing-ai-safety-mentors-and-mentees |
# Injecting some numbers into the AGI debate - by Boaz Barak
I found this article useful context for Boaz Barak's more recent article about [the probability of AI Doom](https://www.lesswrong.com/posts/zB3ukZJqt3pQDw9jz/ai-will-change-the-world-but-won-t-take-it-over-by-playing-3).
In this article, the author (throug... | https://www.lesswrong.com/posts/BaQWrRgu7pjGmBByv/injecting-some-numbers-into-the-agi-debate-by-boaz-barak |
# Against a General Factor of Doom
*Jeffrey Heninger, 22 November 2022*
I was recently reading the results of a survey asking climate experts about their opinions on geoengineering. The results surprised me: “We find that respondents who expect severe global climate change damages and who have little confidence in cu... | https://www.lesswrong.com/posts/Tt83WGrZn6QGcnf2B/against-a-general-factor-of-doom |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.