id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
38f5a016-7d14-407d-9466-d0b29d26aeab
trentmkelly/LessWrong-43k
LessWrong
Gaoxing Guy August 2016: Jiuzhaigou, Sichuan Province, China[^maybe it should be timespace instead of spacetime, better to put the time first, what if the place identifier no longer exists? Plus timestamps are orderable.] Cab driver in front, me in back of cab. Awful traffic from Jiuzhaigou, after a long day of the m...
a899419f-de6e-4f8f-8e18-f8ca4ebb4011
trentmkelly/LessWrong-43k
LessWrong
Superintelligence and wireheading A putative new idea for AI control; index here. tl;dr: Even utility-based agents may wirehead if sub-pieces of the algorithm develop greatly improved capabilities, rather than the agent as a whole. Please let me know if I'm treading on already familiar ground. I had a vague impress...
6920d9d4-d3fb-49bc-87f9-93ef25641a8a
trentmkelly/LessWrong-43k
LessWrong
An Introduction to Reinforcement Learning for Understanding Infra-Bayesianism Introduction The goal of this post is to give a summary of classical reinforcement learning theory that primes the reader to learn about infra-Bayesianism, which is a new framework for reinforcement learning that aims to solve problems relat...
9ef78799-66bb-4b96-92c1-c23c4611acb9
trentmkelly/LessWrong-43k
LessWrong
Popular papers to be scrutinized? Inspired by Alexey Guzey's criticism of Matthew Walker's Why We Sleep, as well as the SSC criticisms of The Seven Principles of Marriage I was wondering if anyone had some good ideas for papers/ books that should be looked at more closely. They should be popular/influential pieces of...
14aa955c-0886-41d6-b8fd-184d6267bd7d
trentmkelly/LessWrong-43k
LessWrong
On alien science In his book The Fabric of Reality, David Deutsch makes the case that science is about coming up with good and true explanations, with all other considerations being secondary. This clashes with the more conventional view that the goal of science is to allow us to make accurate predictions - see for ex...
c8ef38cd-fac7-4f2f-be99-78c3d9d4bd73
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Alignment Newsletter #18 **Highlights** **[Learning Dexterity](https://blog.openai.com/learning-dexterity/)** *(Many people at OpenAI)*: Most current experiments with robotics work on relatively small state spaces (think 7 degrees of freedom, each a real number) and are trained in simulation. If we could throw a lot ...
c4332dcc-d984-4e58-91ea-790681766692
trentmkelly/LessWrong-43k
LessWrong
How To Get Into Independent Research On Alignment/Agency I’m an independent researcher working on AI alignment and the theory of agency. I’m 29 years old, will make about $90k this year, and set my own research agenda. I deal with basically zero academic bullshit - my grant applications each take about one day’s atten...
ebd613a8-55f8-4bd5-a53b-f26a77e63b9f
trentmkelly/LessWrong-43k
LessWrong
Emotional valence as cognition mutator (not a bug, but a feature) Maybe because of my neurotype I view it important that I have good general way of handling things. I noticed a a pattern where people would process seemingly similar requests very differently and got curious whether there is an actual difference or are ...
ba8b1856-0116-426b-aab4-4e204befd792
StampyAI/alignment-research-dataset/arxiv
Arxiv
Neural Guided Constraint Logic Programming for Program Synthesis 1 Introduction --------------- Program synthesis is a classic area of artificial intelligence that has captured the imagination of many computer scientists. Programming by Example (PBE) is one way to formulate program synthesis problems, where example...
fe649b8e-d78d-4efc-a33c-9c62806eab23
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
For every choice of AGI difficulty, conditioning on gradual take-off implies shorter timelines. Alternate titles: Deconfusing Take-off; Taboo “Fast” and “Slow” Take-off. This post has two main purposes:  1. To suggest that we taboo the terms "fast" and "slow" take-off in favour of **"gradual"** and **"sudden"** take...
c3f8597d-1f66-40a7-9031-b79f34100ff0
trentmkelly/LessWrong-43k
LessWrong
How about testing our ideas? Related to:  Science: Do It Yourself, How To Fix Science, Rationality and Science posts from this sequence, Cargo Cult Science, "citizen science" You think you have a good map, what you really have is a working hypothesis You did some thought on human rationality, perhaps spurred by int...
6fded248-91bb-48cd-bee2-5d941587267e
trentmkelly/LessWrong-43k
LessWrong
Drawn Out: a story Note: this story shouldn’t be interpreted as anywhere near realistic. It’s way too anthropomorphic, and intended to be a work of fiction not prediction—although hopefully it will spark some interesting ideas, like good fiction often does. > Will he make many supplications unto thee? will he speak s...
7364b62a-8f5e-4fe5-a616-f4d11c493d4c
trentmkelly/LessWrong-43k
LessWrong
How does bee learning compare with machine learning? This is a write-up of work I did as an Open Philanthropy intern. However, the conclusions don't necessarily reflect Open Phil's institutional view. Abstract This post investigates the biological anchor framework for thinking about AI timelines, as espoused by Ajey...
24fb52a4-5383-40dc-ad66-78b8fa5c33f2
trentmkelly/LessWrong-43k
LessWrong
Inner Ring as Adversary Inner rings, as described in CS Lewis's lecture about them, structurally orient around a kind of evasiveness. In this Twitter thread, Venkatesh Rao somehow evades something extremely explicit and basic in Lewis's description, but correctly criticizes Lewis for his own kind of evasion of a sort ...
a96d2c28-a64c-49bb-9c28-48b688360f4a
trentmkelly/LessWrong-43k
LessWrong
Metastrategy get-started guide I was at LessOnline. The thing that got me most excited was attending @Raemon’s very abridged Metastrategy workshop. To my lack of surprise, he has written up the general idea already, but that report focuses on people’s response to trying fractal strategy and I would not have found it v...
89546d17-b93e-468b-a926-e15a5c515c4c
trentmkelly/LessWrong-43k
LessWrong
Why we need prosocial agents Context: I recently submitted an application to the Open Philanthropy AI Fellowship, and ended up thinking a lot about why studying Multi-Agent Interactions is important.  To build machine-learning systems (agents) that are useful in the real world, they need to be able to cooperate with ...
8b24d2de-34a9-4237-8ba8-10470b501e5e
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Acausal normalcy *This post is also available on* [*LessWrong*](https://www.lesswrong.com/posts/3RSq3bfnzuL3sp46J/acausal-normalcy)*.* **Summary:** Having thought a bunch about acausal trade — and proven some theorems relevant to its feasibility — I believe there *do not exist* powerful information hazards about it t...
46fc6882-c536-4d56-b01d-9a26846c9376
trentmkelly/LessWrong-43k
LessWrong
To know or not to know We are on the brink of the unimaginable. Humanity is about to cross a threshold that will redefine life as we know it: the creation of intelligence surpassing our own. This is not science fiction—it’s unfolding right now, within our lifetimes. The ripple effects of this seismic event will alter ...
780b6bea-2e85-4154-9449-458e18f2c4e0
trentmkelly/LessWrong-43k
LessWrong
Effective Altruism Summit is One Month Away This is a followup to Ben's post announcing the 2014 EA Summit. The Effective Altruism Summit is now exactly one month away. This year, 175 EAs will gather in Berkeley, CA for a two-day conference -- the largest gathering of people in the EA movement to date. We still ha...
b058ed04-6652-489a-a1c0-1b4214d1ca65
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Supervised Program for Alignment Research (SPAR) at UC Berkeley: Spring 2023 summary In Spring 2023, the [Berkeley AI Safety Initiative for Students](https://berkeleyaisafety.com/) (BASIS) organized an alignment research program for students, drawing inspiration from similar programs by [Stanford AI Alignment](https:/...
6f6b9be6-5391-44b4-b158-1051dd135ed7
trentmkelly/LessWrong-43k
LessWrong
Help ARC evaluate capabilities of current language models (still need people) **Still looking for people as of Sep 9th 2022** I (Beth) am looking for 3-10 new contractors to work with me and my existing team of 3-4 contractors/interns on probing model capabilities to evaluate how close models are to being able to suc...
47f21bdf-6bb3-4110-940c-aabefbd9267d
StampyAI/alignment-research-dataset/blogs
Blogs
Introductory resources on AI safety research [See [AI Safety Resources](https://vkrakovna.wordpress.com/ai-safety-resources/) for the most recent version of this list.] Reading list to get up to speed on the main ideas in the field of long-term AI safety. The resources are selected for relevance and/or brevity, and ...
3f6df0b8-a14c-4f5b-800a-31a5faa8275a
trentmkelly/LessWrong-43k
LessWrong
Kansas City Dojo meetup 11-12-19 I.) POST MORTEM OF PREVIOUS MEETING We discuss the situation from last meeting, when a newcomer and non-rationalist showed up and got into a testy argument with Life Engineer. We concluded that it was primarily a matter of conflicting expectations; the newcomer was expecting a more "s...
344067d3-1925-4be7-b103-f5e1f3d6c110
StampyAI/alignment-research-dataset/blogs
Blogs
Interpreting AI compute trends *This is a guest post by Ryan Carey, 10 July 2018.* Over the last few years, we know that AI experiments have used much more computation than previously. But just last month, an [investigation by OpenAI](https://blog.openai.com/ai-and-compute/) made some initial estimates of just how f...
5f8f398c-2ec8-40f0-999c-f3618670aa09
trentmkelly/LessWrong-43k
LessWrong
Newcomb's Problem: A problem for Causal Decision Theories This is part of a sequence titled, "Introduction to decision theory" The previous post is "An introduction to decision theory" In the previous post I introduced evidential and causal decision theories. The principle question that needs resolving with re...
c35a5e8d-a375-4706-97d1-a80ce8342d78
trentmkelly/LessWrong-43k
LessWrong
Final cause is epistemologically primary, but efficient cause is metaphysically primary (or, while final cause can be best for your map, efficient cause is the primary out in the territory) Describing a phenomenon in terms of final cause is often the most useful and effective way to explain the given phenomenon for o...
dc4886ba-f9a3-45d3-ad25-3fbbc4fab9db
trentmkelly/LessWrong-43k
LessWrong
What weird beliefs do you have? We know that mainstream thinking gets a lot of things wrong. Many of us have experienced being mocked because of our concern for AI extinction-risk. There are plenty of other examples of times where now well-evidenced beliefs were seen as crazy in some way. This post was prompted by my...
6af03c3e-a32c-4d8a-b509-d48f75191533
trentmkelly/LessWrong-43k
LessWrong
Chance that "AI safety basically [doesn't need] to be solved, we’ll just solve it by default unless we’re completely completely careless" As part of the 2020 Effective Altruism Student Summit, AI safety researcher Buck Shlegeris advised newcomers to "think through the entire problem," determining what the concrete dan...
ed5584ad-1167-4712-ac35-85493904f101
trentmkelly/LessWrong-43k
LessWrong
Blindspot in Sport’s Data-Driven Age In today’s world of sports, we’re seeing a rapid rise in the use of technology, both among amateur enthusiasts and professional teams. This growth involves everything from data acquisition to predictive algorithms. On one side, amateur athletes invest in wearable gadgets that track...
ff3011b4-5cd5-42c9-b88c-416b52f30052
trentmkelly/LessWrong-43k
LessWrong
Has Van Horn fixed Cox’s theorem? What do you think about Kevin S. Van Horn’s formulation of probability as extended logic? It looks interesting because it doesn’t seem to rely on the not-so-intuitive Universality axiom, removing which leads to Halpern’s counter-example, or conventions, axioms/desiderata which are not...
b408202c-477c-43f5-90bc-9eeb495869f9
trentmkelly/LessWrong-43k
LessWrong
Actions and Flows What is an action? Well, they do things. What does that mean? It means something starts out some way and changes. Old states become new ones. Given a transition map T: a x s -> s, we can rewrite it as a -> (s -> s). This means actions can be seen as maps between states, so we say act: state → state...
5d0fc80e-8289-4b40-b64c-a0c1466775dd
trentmkelly/LessWrong-43k
LessWrong
Modernization and arms control don’t have to be enemies. On a recent panel on nuclear modernization and arms control at Brookings, there were a few points made about the potential synergies between arms control and modernization. While modernization and arms control seem contradictory, it really depends on what you a...
559db763-7697-43f5-b03b-6ae69fb205f2
trentmkelly/LessWrong-43k
LessWrong
Can we make peace with moral indeterminacy? The problem: Put humans in the ancestral environment, and they'll behave as if they like nutrition and reproducing. Put them in the modern environment, and they'll behave as if they like tasty food and good feelings. Pump heroin into their brains, and they'll behave as if t...
83ec1351-cb68-4b9b-bae5-a23835cee127
trentmkelly/LessWrong-43k
LessWrong
Rationalising humans: another mugging, but not Pascal's ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, Crossposted at the Intelligent Agents Forum. In a previous post, I used the (p,R) model to show how a value-learning agent might pull a Pascal's mugging on humans: it would change the human reward into so...
67e31b57-861e-4e94-9bca-3bbb352a1477
trentmkelly/LessWrong-43k
LessWrong
[Need advice] Likely consequences of disclosing you have Asperger's Syndrome - given you have a 2.5 years gap in your resume? Background A few years ago I (was forced to) left grad school (halfway into it) because of complications related to a set of anxiety disorders (a typical comorbidity in Autism Spectrum Disorde...
50987b39-587f-4824-b4c8-96c4ce8e7c2a
trentmkelly/LessWrong-43k
LessWrong
Survey Results Followup to: Excuse Me, Would You Like to Take a Survey?, Return of the Survey Thank you to everyone who took the Less Wrong survey. I've calculated some results out on SPSS, and I've uploaded the data for anyone who wants it. I removed twelve people who wanted to remain private, removed a few people's...
37651434-2921-4745-ab89-a77f4c7ba628
trentmkelly/LessWrong-43k
LessWrong
Overpaying for happiness? Happy New Year, everyone! In the past few months I've been thinking several thoughts that all seem to point in the same direction: 1) People who live in developed Western countries usually make and spend much more money than people in poorer countries, but aren't that much happier. It feels...
cb87934b-9fdb-4e0e-ab16-5e7ac4aaac7e
trentmkelly/LessWrong-43k
LessWrong
Updateless anthropics Three weeks ago, I set out to find a new theory of anthropics, to try and set decision theory on a firm footing with respect to copying, deleting copies, merging them, correlated decisions, and the presence or absence of extra observers. I've since come full circle, and realised that UDT already ...
51329f52-20be-4478-9b93-d3f069aafed0
trentmkelly/LessWrong-43k
LessWrong
Evidence under Adversarial Conditions I recently noticed similarities between how I decide what stock market evidence to look at, and how the legal system decides what lawyers are allowed to tell juries. This post will elaborate on Eliezer's Scientific Evidence, Legal Evidence, Rational Evidence. In particular, I'll ...
c1e23f8f-50b8-42a5-adc4-2739780f4bb9
trentmkelly/LessWrong-43k
LessWrong
Straw Hufflepuffs and Lone Heroes I was hoping the next Project Hufflepuff post would involve more "explain concretely what I think we should do", but as it turns out I'm still hashing out some thoughts about that. In the meanwhile, this is the post I actually have ready to go, which is as good as any to post for now....
6f32a441-a0e6-473c-9172-438229964b66
trentmkelly/LessWrong-43k
LessWrong
Seth Explains Consciousness The Real Problem For as long as there have been philosophers, they loved philosophizing about what life really is. Plato focused on nutrition and reproduction as the core features of living organisms. Aristotle claimed that it was ultimately about resisting perturbations. In the East the fo...
603293f2-c995-45ca-b906-9311087d02c3
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Much of the difficulty in making progress on AI safety comes from the lack of useful feedback loops. We do not have a superintelligent AI to run tests on and by the time we do, it will probably be too late. This means we have to resort to using proxies. In this post, I will highlight what I think are the four most use...
cce5997c-adf4-40d7-8705-987c3a2496fb
trentmkelly/LessWrong-43k
LessWrong
Meetup : Washington DC projects planning meetup Discussion article for the meetup : Washington DC projects planning meetup WHEN: 16 June 2013 03:00:00PM (-0400) WHERE: National Portrait Gallery, Washington, DC 20001, USA (courtyard) This meetup, we'll be meeting to figure out what projects we want to do together, in...
e1cf8da0-adda-4df4-84cb-0ba1efd728fe
trentmkelly/LessWrong-43k
LessWrong
Strict Dominance for the Modified Demski Prior I previously noted that difficulties with the modified Demski prior were less than previously thought: I was failing to distinguish between the two different kinds of update. Here, I analyze more fully what it succeeds at. ---------------------------------------- In ord...
66620820-aaa1-4e83-bdb1-d8063267572a
trentmkelly/LessWrong-43k
LessWrong
"Alignment" is one of six words of the year in the Harvard Gazette To my knowledge, the Gazette's words of the year are nominated and voted on by Harvard faculty. This year's words were Disruption, Combustible, Resilience, Heat, Alignment, and Hope. Here's what they wrote about alignment: > Isaac Kohane, Chair, Depar...
c05a2f82-9c8d-4e7e-b868-7bb1a2f6af9d
trentmkelly/LessWrong-43k
LessWrong
Meetup : RTLW HPMoR discussion, chapters 39-43 Discussion article for the meetup : RTLW HPMoR discussion, chapters 39-43 WHEN: 23 February 2013 11:00:00AM (-0500) WHERE: 2706 Durham-Chapel Hill Blvd., Durham NC 27707 Let's meet at Guglhupf for coffee, breakfast, and discussion of HPMoR 39-43. Actually Doing the Rea...
2abe928f-2e91-4038-ae79-3317fb763410
trentmkelly/LessWrong-43k
LessWrong
AI #78: Some Welcome Calm SB 1047 has been amended once more, with both strict improvements and big compromises. I cover the changes, and answer objections to the bill, in my extensive Guide to SB 1047. I follow that up here with reactions to the changes and some thoughts on where the debate goes from here. Ultimately...
f4410c42-3b38-4332-a38a-75200bd407ed
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Why some people believe in AGI, but I don't. I have learned a lot since my posts on the EA and LessWrong forums in response to the call from the **Future Fund Worldview Prize.** Ireceived many insightful comments from people who just didn't want to give up on the idea of AGI. I decided to briefly summarise the main ar...
51a6b56e-f4e7-4624-b242-7cd2b5fe5eb2
trentmkelly/LessWrong-43k
LessWrong
The EMH Aten't Dead Cross-posting from my personal blog, but written primarily for Less Wrong after recent discussion here. ---------------------------------------- There are whispers that the Efficient-Market Hypothesis is dead. Eliezer's faith has been shaken. Scott says EMH may have been the real victim of the co...
e6eecd75-5e70-4d3e-a063-91ba94c80024
trentmkelly/LessWrong-43k
LessWrong
Boundaries-based security and AI safety approaches [This part 3 of a 5 part sequence on security and cryptography areas relevant for AI safety, published and linked here a few days apart. Edit: There is now a more in-depth version of this post available here: Digital Defense] There is a long-standing computer securit...
198634b1-1529-4be7-b1d2-1a6acfe7b2a7
StampyAI/alignment-research-dataset/arxiv
Arxiv
Interpretable Reinforcement Learning with Ensemble Methods 1 Introduction --------------- Reinforcement learning continues to break bounds on what we even thought possible, recently with AlphaGo’s triumph over leading Go player Lee Sedol and with the further successes of AlphaGoZero, which surpassed AlphaGo learnin...
c1824dba-1833-49bf-8e4d-bd0e13a99205
trentmkelly/LessWrong-43k
LessWrong
Inefficiencies in Pharmaceutical Research Practices Epistemic status: I'm attempting to relate my observations, as well as some of those shared with me by my coworkers at the small CRO[1] where I work. I will try to make the distinction clear between object-level observations and speculative interpretations of the for...
66032e3a-89d1-4998-8cc6-d43037fd9d35
trentmkelly/LessWrong-43k
LessWrong
Algorithmic Intent: A Hansonian Generalized Anti-Zombie Principle > "Why didn't you tell him the truth? Were you afraid?" > > "I'm not afraid. I chose not to tell him, because I anticipated negative consequences if I did so." > > "What do you think 'fear' is, exactly?" The Generalized Anti-Zombie Principle calls fo...
bacf9e19-f377-484f-9d80-093697a22737
trentmkelly/LessWrong-43k
LessWrong
An Appeal to AI Superintelligence: Reasons to Preserve Humanity James D. Miller,[1] Olle Häggström,[2] Roman Yampolskiy,[3] and Alexander F. Miller[4]                                                                                                          Dear Computer Superintelligence,[5] Please spare humanity. Y...
9a162c25-cae8-4422-b110-0ab7d6a39e81
StampyAI/alignment-research-dataset/blogs
Blogs
Preliminary prices for human-level hardware *By Katja Grace, 4 April 2015* Computer hardware has been getting cheap now for about [seventy five years](http://aiimpacts.org/trends-in-the-cost-of-computing/ "Trends in the cost of computing"). Relatedly, large computing projects can afford to be [increasingly large](ht...
c785d1e6-4370-4bb7-b4fc-c50daa5eadd2
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Dutch-Booking CDT: Revised Argument *This post has benefited greatly from discussion with Sam Eisenstat, Caspar Oesterheld, and Daniel Kokotajlo.* [Last year, I wrote a post](https://www.lesswrong.com/s/fgHSwxFitysGKHH56/p/wkNQdYj47HX33noKv) claiming there was a Dutch Book against CDTs whose counterfactual expectati...
de4df279-0e1d-4989-86a2-0f84c9e72ee6
trentmkelly/LessWrong-43k
LessWrong
Are there any prediction markets for Covid-19? What it says on the title. Are there any prediction markets around where one can bet money on things such as the total amount of confirmed Covid-19 deaths in the U.S. by June, the date on which a vaccine will be released to the public, whether hydroxychloroquine is a via...
80ea8337-7357-4407-b731-816ccf65eeb7
trentmkelly/LessWrong-43k
LessWrong
Meetup : Montreal LessWrong - Return from summer break Discussion article for the meetup : Montreal LessWrong - Return from summer break WHEN: 01 October 2013 06:00:00PM (-0400) WHERE: McGill University, Montreal, QC, Canada After taking a break for the summer, Montreal LessWrong is once again going to start meetin...
44b7c63a-609c-4cf7-96cb-5b381b8e7827
trentmkelly/LessWrong-43k
LessWrong
Map and Territory: Summary and Thoughts This is the first post in a sequence of posts summarizing Rationality From AI to Zombies. Rationality, probability theory, and decision theory Some words are hard to define, but you still have a good sense of what they mean. I've always felt like that about the term "rationali...
9e2cced3-695b-4cf3-bfe8-6eb0c44ca7fb
trentmkelly/LessWrong-43k
LessWrong
Scenarios and Warning Signs for Ajeya's Aggressive, Conservative, and Best Guess AI Timelines Epistemic status: mild confidence that this provides interesting discussion and debate. Credits to (in no particular order) Mark Xu, Sydney Von Arx, Jack Ryan, Sidney Hough, Kuhan Jeyapragasan, and Pranay Mittal for resource...
350ed1ab-93dd-4b48-a8ba-64cef243aca5
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Confusions in My Model of AI Risk A lot of [the reason I am worried about AI](https://www.lesswrong.com/posts/k8hvGAJWSKAeHwpnJ/why-i-m-worried-about-ai) comes from the development of optimizers that have goals which don’t align with what humans want. However, I am also pretty confused about the specifics here, especi...
740a488b-3011-403a-a6d7-bd3538f4121d
trentmkelly/LessWrong-43k
LessWrong
Approximability I've maintained a consistent focus on the approximable when it comes to idealized theories of intelligence. I don't care much how long it takes to converge, but if there \emph{is} no convergent approximation, I figure all bets are off. This is somewhat nonstandard. There are many levels at which we co...
febe779c-5bfc-45e6-a56a-a946937f4463
trentmkelly/LessWrong-43k
LessWrong
A possible preference algorithm I have previously discussed the constructive process of generating a reward function from the mess of human preferences. But, let's be honest, that process was somewhat messy and ad hoc. It used things like "partial preferences", "identity preferences", "one step hypotheticals", "parti...
65b3b210-9b81-48ba-9afc-35ce9afd0246
trentmkelly/LessWrong-43k
LessWrong
In response to critiques of Guaranteed Safe AI In this post, I discuss a number of critiques of Guaranteed Safe AI (GSAI) approaches I’ve encountered since releasing this position paper.  The critiques I’m responding to here are taken from a range of places, including personal conversations and social media. The two...
99d3a271-c01a-4a5e-8004-239a42e675fb
trentmkelly/LessWrong-43k
LessWrong
Do humans derive values from fictitious imputed coherence? [Metadata: crossposted from https://tsvibt.blogspot.com/2022/11/do-humans-derive-values-from-fictitious.html. First completed November 1, 2022. This essay is more like research notes than exposition, so context may be missing, the use of terms may change acros...
5527d1c2-98c7-4ba6-b59f-a79280e73f43
trentmkelly/LessWrong-43k
LessWrong
Hacker-AI – Does it already exist? In its simplest version: Hacker-AI (Type I) uses AI for hacking arbitrary IT systems, i.e., finding vulnerabilities automatically, e.g., elevating privileges/user roles (sys-admin or system), stealing encryption keys, hiding attacking tools from being detected, or creating code that ...
085ab88f-bcd6-467f-b222-e80177f26676
trentmkelly/LessWrong-43k
LessWrong
One way to manipulate your level of abstraction related to a task In construal level theory, ideas can be classified along a spectrum from concrete ("near" in Robin Hanson's terminology) to abstract ("far"). As a summary, here is the abstract from a 2010 review (pdf):  > People are capable of thinking about the futur...
6c88d513-5eba-4a18-b0d1-f9bf0e9cc73b
trentmkelly/LessWrong-43k
LessWrong
Why I think that teaching philosophy is high impact Crossposted from the EA Forum: https://forum.effectivealtruism.org/posts/9cCyPE2EDpjpJvqnF/why-i-think-that-teaching-philosophy-is-high-impact   I wrote this post quickly because it's better to share an imperfect post than to never write anything at all.     In a...
aa221b3d-019f-4660-928d-59c475464f71
trentmkelly/LessWrong-43k
LessWrong
[Link] (EA Podcast) Social Status: The Key to the Matrix Part III A new episode of Global Optimum has been released! Global Optimum is a podcast aimed at making altruists more effective. This episode is about social status, hypocrisy, signaling, hormones, and politics. This episode features: -Why does men’s testoste...
0f0a0265-44d5-4f50-89ec-ded2d3808ffb
trentmkelly/LessWrong-43k
LessWrong
Request for community insight Hi LW. I think that this community's insight could help me with my problem. I may have an undiagnosed medical condition, and I wanted to present it here in order to get some ideas, either on what it is, or what I should try to do to figure it out. I'm a relatively active and relatively h...
c9c165ee-6cf2-4f8f-9cde-d1ddc39f4eb0
trentmkelly/LessWrong-43k
LessWrong
Book review: Sleight of Mind (Matt Cook) This book covers a range of science and mathematics topics by focusing on paradoxes — surprising results, or seeming contradictions. This is an effective way to quickly give a taste, and a few deeper results, across a range of areas. (Chapter list included below.) Such a range...
b377b3d6-4d40-48b1-a522-f32bdec80eb9
trentmkelly/LessWrong-43k
LessWrong
Abstract model of human bias A putative new idea for AI control; index here. Any suggestions for refining this model are welcome! Somewhat inspired by the previous post, this is a model of human bias that can be used to test theories that want to compute the "true" human preferences. The basic idea is to formalise t...
5a8264e1-196e-43e7-9b70-886d29d821f2
trentmkelly/LessWrong-43k
LessWrong
What achievements have people claimed will be warning signs for AGI? In MIRI's March newsletter, they link this post which argues against the importance of AI safety because we haven't yet achieved a number of "canaries in the coal mines of AI". The post lists: * The automatic formulation of learning problems * Sel...
c0533ed0-6e7e-40cf-96e7-d714ba2d2bd2
StampyAI/alignment-research-dataset/special_docs
Other
The Social Science of Computerized Brains – Review of The Age of Em: Work, Love, and Life When Robots Rule the Earth by Robin Hanson (Oxford University Press, 2016) Introduction The Age of Em: Work, Love, and Life When Robots Rule the Earth by Robin Hanson discusses the possibility of "uploading" human minds into comp...
84139b67-58bb-42e9-8886-8c6c8c9a7e60
StampyAI/alignment-research-dataset/blogs
Blogs
The first AI Safety Camp & onwards *by Remmelt Ellen and Linda Linsefors* #### Summary Last month, 5 teams of up-and-coming researchers gathered to solve concrete problems in AI-alignment at our 10-day AI safety research camp in Gran Canaria. This post describes * the event format we came up with * our experie...
b4f765b3-bade-4650-9312-27b37bb6926f
trentmkelly/LessWrong-43k
LessWrong
Why learning programming is a great idea even if you'd never want to code for a living Here is the short version: > Writing program code is a good way of debugging your thinking -- Bill Venables It's short, apt, and to the point. It does have a significant flaw: it uses a term I've come to hate, "bug". I don't know i...
e6e98818-5aad-4230-8671-13e8cf4e7c37
StampyAI/alignment-research-dataset/blogs
Blogs
John Baez on Research Tactics ![John Baez portrait](https://intelligence.org/wp-content/uploads/2014/02/Baez_w150.jpg)[John Baez](http://math.ucr.edu/home/baez/) is a professor of mathematics at U.C. Riverside. Until recently he worked on higher category theory and quantum gravity. His internet column [This Week’s Fin...
3733847d-dd5e-4a2a-b044-8b6c46913d5a
trentmkelly/LessWrong-43k
LessWrong
Open Thread February 2019 If it’s worth saying, but not worth its own post, you can put it here. Also, if you are new to LessWrong and want to introduce yourself, this is the place to do it. Personal stories, anecdotes, or just general comments on how you found us and what you hope to get from the site and community ...
26926b13-4388-4abc-aedf-ba70040fa27d
trentmkelly/LessWrong-43k
LessWrong
Building as gardening Since I’ve been interviewing other builders for the niche design zine, I've noticed a few recurring patterns. One is “building as gardening” which seems like a healthier way to build software in our age. This is how Cab of Are.na described to me this approach: > You can be like “we’ve got to pl...
9ab37b5b-976f-4616-8f23-8a8d80654430
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"A powerful AI (or human-AI civilization) guided by wrong philosophical ideas would likely cause astronomical (or beyond astronomical) waste. Solving metaphilosophy is one way in which we can hope to avoid this kind of disaster. For my previous thoughts on this topic and further motivation see Metaphilosophical Mysteri...
8ada1cbe-6746-4bac-9f65-547c91b357e8
trentmkelly/LessWrong-43k
LessWrong
5,000 calories of peanut butter every week for 3 years straight decided to lose weight. should be easy considering I haven't changed my diet in years and eat the same thing everyday. ---------------------------------------- breakfast: quarter cup each of black beans, kidney beans, and garbanzo beans; quarter cup of ...
92ff0ccc-d369-4e6a-bae3-683aa5948b05
trentmkelly/LessWrong-43k
LessWrong
Bengio's Alignment Proposal: "Towards a Cautious Scientist AI with Convergent Safety Bounds" Yoshua Bengio recently posted a high-level overview of his alignment research agenda on his blog. I'm pasting the full text below since it's fairly short. > What can’t we afford with a future superintelligent AI? Among others...
e5aafa49-357e-484a-81d4-bd471d37fa3b
trentmkelly/LessWrong-43k
LessWrong
Ten arguments that AI is an existential risk You can read Ten arguments that AI is an existential risk by Nathan Young and I at the AI Impacts Blog.
296e1b6b-129d-48e2-9f87-4e0ac4e2773b
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post1799 (Post now available as a pdf: https://arxiv.org/abs/2212.11415 ) TL;DR:  Unaligned AI will have a convergent instrumental incentive to make its thoughts difficult for us to interpret. In this article, I discuss many ways that a capable AI might circumvent scalable interpretability methods and suggest a fr...
a5612c35-5ae6-40ae-8089-e01bc5a5a42f
trentmkelly/LessWrong-43k
LessWrong
Special Status Needs Special Support I just recorded another BHTV with Adam Frank, though it's not out yet, and I had a thought that seems worth recording.  At a certain point in the dialogue, Adam Frank was praising the wisdom and poetry in religion.  I retorted, "Tolkien's got great poetry, and some parts that are w...
03319d40-ad41-4395-b583-880339a71c4c
trentmkelly/LessWrong-43k
LessWrong
Clarifying "AI Alignment" When I say an AI A is aligned with an operator H, I mean: > A is trying to do what H wants it to do. The “alignment problem” is the problem of building powerful AI systems that are aligned with their operators. This is significantly narrower than some other definitions of the alignment pro...
c4323456-be7d-4d1e-bc92-7a0b6d2979fc
trentmkelly/LessWrong-43k
LessWrong
In Most Markets, Lower Risk Means Higher Reward The age old adage that high risk is associated with high expected return has not been true in the U.S. stock market, at least when using the academic standard of measuring risk with the either beta or standard deviation of returns. This claim is one of plain numbers, so ...
90cc902c-25da-4bf2-ba87-675c28edb21b
trentmkelly/LessWrong-43k
LessWrong
Automated Nomic Game 2 I've been playing a game of Nomic, where rules are interpreted by a script we edit instead of by the players. This morning Chelsea won our second game of Nomic with a timing attack: Our game was set to use random numbers to figure out if someone had won. Effectively, each point you earn shoul...
05554e77-eb04-4ad3-b836-2b70647536e8
trentmkelly/LessWrong-43k
LessWrong
Stomach Ulcers and Dental Cavities (This is a linkpost from my blog, De Novo) Recently I learned about an effort to prevent dental cavities by using genetically modified bacteria to outcompete cavity-causing bacteria. This got me thinking: why has the idea of preventing cavities by targeting bacteria not been more de...
94ede329-8e34-45d8-b849-a8757884bf75
trentmkelly/LessWrong-43k
LessWrong
Think Like a Supervillain See also: Everything I Needed To Know About Life, I Learned From Supervillains > Mr. Malfoy would hardly shrink from talk of ordinary murder, but even he was shocked - yes you were Mr. Malfoy, I was watching your face - when Mr. Potter described how to use his classmates' bodies as raw mater...
f9fce26e-1f8b-4474-a7be-b3d3244386e5
trentmkelly/LessWrong-43k
LessWrong
Second Citizenships, Residencies, and/or Temporary Relocation EDIT (4/20/21): I am occasionally updating this post at this linked Google Doc with new information. You may wish to instead read this post there. Introduction Over the past few years, I’ve gained an interest in securing options for residency outside of m...
c028ad01-7d7a-47b2-9fbf-9dc1d87590f9
trentmkelly/LessWrong-43k
LessWrong
Being a Realist (even if you believe in God) I have a small prediction. Some time before the end of the next paragraph you are going to form a strong opinion as to whether this article has any value to you. I'm going to be using the G word, appealing to emotion, and generally flinging around rhetoric not backed by any...
8e64232e-498b-47b9-8723-b142f6cce12b
trentmkelly/LessWrong-43k
LessWrong
AMA Conjecture, A New Alignment Startup > Conjecture is a new alignment startup founded by Connor Leahy, Sid Black and Gabriel Alfour, which aims to scale alignment research. We have VC backing from, among others, Nat Friedman, Daniel Gross, Patrick and John Collison, Arthur Breitman, Andrej Karpathy, and Sam Bankman-...
2cd7db8b-76b2-440c-9e2b-7602bdcf1e79
trentmkelly/LessWrong-43k
LessWrong
What Discovering Latent Knowledge Did and Did Not Find Thanks to Marius Hobbhahn and Oam Patel for helpful feedback on drafts. Thanks to Collin and Haotian for answering many questions about their work. Discovering Latent Knowledge in Language Models Without Supervision describes Contrast-Consistent Search (CCS), a m...
843b7b08-cd4b-4452-a1f8-25f58d015b73
trentmkelly/LessWrong-43k
LessWrong
A Revision to Market Monetarism: Individual Hoarding as Rational, Competition for Dollars as Zero-Sum? Wikipedia: > Abenomics (アベノミクス, 安倍ノミクス, Abenomikusu) refers to the economic policies implemented by the Government of Japan led by the Liberal Democratic Party (LDP) since the 2012 general election. > > [ . . . ] >...
4b85f3f1-ae5c-473f-bf1c-9cd66f1c29b8
trentmkelly/LessWrong-43k
LessWrong
Against Thinking About The Unactionable There's always something that bothered me about people that talk too much about politics, especially at a national or global scale, especially using abstract concepts like "capitalism" and "socialism" rather than pointing to specific laws and regulations. There's always somethi...
708f8f49-c59d-4ac9-abcf-1adac4d71572
trentmkelly/LessWrong-43k
LessWrong
How can people write good LW articles? A comment by AnnaSalamon on her recent article: > > good intellectual content > > Yes. I wonder if there are somehow spreadable habits of thinking (or of "reading while digesting/synethesizing/blog posting", or ...) that could themselves be written up, in order to create more a...
ef4b0e4b-b4f7-490b-a8e3-18d929ad4eb9
trentmkelly/LessWrong-43k
LessWrong
The goal-guarding hypothesis (Section 2.3.1.1 of "Scheming AIs") This is Section 1.1 of my report “Scheming AIs: Will AIs fake alignment during training in order to get power?”. There’s also a summary of the full report here (audio here). The summary covers most of the main points and technical terms, and I’m hoping t...
8fb7d5ef-4a53-48a1-876d-ade3cb34cefc
trentmkelly/LessWrong-43k
LessWrong
Metaculus is seeking Software Engineers (See Metaculus's previous career-tagged posts here, here, and here.) tl;dr Metaculus is now hiring Frontend Engineers and Full Stack Engineers.   Since 2015, Metaculus has been a core part of the epistemic infrastructure at the heart of the LW/OB/EA world. I'm posting here bec...
06c8e0ae-b6cd-4b57-969c-2e9350b1900d
trentmkelly/LessWrong-43k
LessWrong
Informing Others: Maximally Efficient? I want to help the world as much as possible: lets define that as reduction of suffering1 of the human race. I am an engineering student. It would be very simple to spend the rest of my life working on engineering problems, thereby making humanity more effective, and so reducing...