4,318 karma · joined April 27, 2013
> What has Eliezer Yudkowsky produced, in terms of scientific output? I've read plenty of his lesswrong's writing on the idea of rationalist, as well as plenty more on the AI alignment side (obviously including the famous basilisk). Can you point me to some results from the last 2 decades of his works? Either from him or his institute. That helps to calibrate what his expectation for the AI safety field.
And this was the answer it wrote - I've prefaced each paragraph with "GPT-4>" to indicate the source. It is not fact-checked, and indeed, it has at least one error: The paper GPT-4 claimed to introduce "the concept of Logical Induction" actually introduces "logical induction criterion" and speaks about evaluating algorithms for Logical Induction. Close, but wrong. But overall it is an interesting response - just not a carefully considered one by how we currently would describe logical reasoning, etc.
GPT-4> Eliezer Yudkowsky is a well-known researcher, writer, and advocate for AI safety and rationality. He has contributed significantly to the fields of AI alignment and rational thinking through his work at the Machine Intelligence Research Institute (MIRI), formerly known as the Singularity Institute for Artificial Intelligence (SIAI), and through his writings on LessWrong. Here are some notable outputs from Yudkowsky and MIRI:
GPT-4> 1. Coherent Extrapolated Volition (CEV): Yudkowsky introduced the concept of CEV as a way to describe the ideal values and preferences of humanity that an AI should pursue. The CEV aims to find the values that would emerge if humans had the opportunity to reflect on their goals and beliefs more thoroughly and coherently.
GPT-4> 2. Timeless Decision Theory (TDT): Yudkowsky developed TDT as an alternative to standard decision theories such as Causal Decision Theory (CDT) and Evidential Decision Theory (EDT). TDT aims to better address problems like Newcomb's Problem and counterfactual mugging by considering the logical implications of an agent's decision, rather than just the causal consequences.
GPT-4> 3. Research on Logical Induction: MIRI researchers, including Scott Garrabrant and others, published a paper in 2016 that introduced the concept of logical induction. This work is an attempt to formalize the process of reasoning under logical uncertainty and updating beliefs about mathematical statements.
GPT-4> 4. Yudkowsky's sequence of essays on rationality: Yudkowsky has written extensively on various topics related to rationality, including epistemic rationality, instrumental rationality, and biases. These writings are accessible on LessWrong and have been compiled into the book "Rationality: From AI to Zombies."
GPT-4> 5. AI safety research: MIRI has produced a range of research on AI safety topics, including value alignment, decision theory, and robustness. Much of this work has been foundational in shaping the AI alignment research landscape.
GPT-4> While Yudkowsky's work might not have the same breadth of peer-reviewed publications as some other AI researchers, his contributions have been influential in the AI alignment and rationality communities. His ideas have helped shape the ongoing discussion about AI safety and the development of effective strategies for managing advanced AI systems.
Even the abbreviation has a history. And a couple Unicode symbols: and ⳩ (weirdly, one displays in HN and the other does not.)
Too true. I’m from Canada where we worry nurses will go south of the border, but was recently in London at an NHS hospital and was talking about this with a nurse. Pay for nurses in Canada is on par with a Jr Developer here. (Still low for the hours and effort required!) By comparison, nurses in the UK make about as much as an intern or admin assistant. After all, single payer means there isn’t much pressure to raise wages from other sectors, you’ve a monopoly on nursing so you can set the wages at whatever level keeps staffing adequate, which isn’t the same thing as paying a living wage…
> To aid refactoring, double-quote characters can be used inside embedded expressions without escaping them as \".
I would expect for consistency that \"{expression}\" outputs an expression and \"\{expression}\" would not.
Basically, amyloids have been known for centuries as e.g. white stuff that collects in spleens and recently research on dementia has turned to amyloids as one of the precursors of one of the types of Alzheimer’s. It’s not yet clear if every case has amyloid involvement or if clearing amyloids can resolve the condition without any side effects yet, in particular there’s been reported brain shrinkage.
https://en.m.wikipedia.org/wiki/Amyloid_beta talks about the thing floating around your body in more detail.
I can’t say I’m an expert at any of this though. I’ve mostly looked at this from an AA amyloidosis perspective rather than an Altzheimer’s one.
We actually already have a test for the presence of amyloids - SAA blood tests commonly used to diagnose infection in horses and cats. Other creatures, including humans, can also measure C Reactive Protein, but both CRP and SAA are measures of stress or inflammation associated with bacterial infection. Your body routinely gets rid of amyloids it might produce, which is part of why it’s said people with AA amyloidosis recover when the underlying cause is treated, I think. But it’s unclear how well we can recover long-term and amyloids and biofilms are still active research areas.
All I can say is that for my amyloidosis-like symptoms, taking vitamin C regularly helps my thinking and the occasional high dose of vitamin A appears to temporarily reverse some of the amyloid-induced nerve damage. But I should also clarify I’m still awaiting test results to determine if I have amyloids and also my condition improves with antibiotics which means vitamin A and C might be interfering with bacteria (where biofilm = amyloids) rather than with the amyloid fibrils on their own.
On Windows a similar app was Jing. https://en.m.wikipedia.org/wiki/Jing_(software) but the Skitch app could also really easily add text, and had a neat way of dragging and dropping the tab containing the filename to save a file to a particular location or send it somewhere that you rarely see today. It wasn’t intuitive but it was easy once you knew how, like dragging the file icon from the window title bar (a hidden feature of only some document-based macOS apps).
Turning on zooming likely refers to https://support.apple.com/en-ca/HT210978 which was a feature added to Snow Leopard (maybe) where you could hold the control key and use the mouse wheel or Magic Mouse to zoom in.
I was going to stop by to post about the feature, documented on the above page, under “Play audio on a HomePod using Control Center”
If it showed up anywhere in future, it would be there, I think, in the list of AirPlay devices.
I can say I’ve had the opposite problem, where I’m on Spotify on one device and, while playing, would want to use a second device to preview tracks. But the Spotify model only lets you play in one location at a time. I do like the Spotify model most of the time, though.
Feels like there’s a missing password prompt there. And maybe confirmation on a second device, if you have one.
For a library that generates code, that’s a special case, as the code it generates must target a particular language version. You have three choices: 1. Upstream a fix as you propose; 2. Side-by-side install both TS 4.6 and TS 4.7 using workspaces or sub-projects and have some of your code compile with 4.6 and then link the results or 3. Find a replacement that is updated to 4.7. For example, https://github.com/stephenh/ts-proto has 4.7 support listed in its readme.
Chrome uses HTML5 search inputs: https://martin-thoma.com/search-engine-autodiscovery/
But yes, I do miss reading magazines for the ads ;-)
Part of its source material is from a 2002 publication: https://www.nps.gov/subjects/nationalregister/upload/NRB46_S... (8mb PDF, see page 59 of the PDF, with printed page number of 49.)
The blog post is (2018) and the original article it republishes is (1996) according to Google.
You can also find it on Archive.org at https://web.archive.org/web/20220504061157/https://dl.acm.or...
This really is an area that needs more innovation, long term.
Oh, I also sometimes embed videos using iframes into Miro boards and then use that for note taking. But it’s a little tricky when you want to attach notes to particular timestamps in the video. I think maybe you can embed Descript shares as iframes into Miro but I haven’t tried it yet.
Now, the obvious answer then is that for most online stores, somebody picks a product classification at random or relies on a third-party service that matches a product description to a classification. Most of the time, getting it wrong has few consequences unless you’ve been informed of the problematic item and the same government sees a misclassification more than once, or wants to make a point with your shipment vs others. After all, saying others get away with it doesn’t absolve your need to try and get declarations and duties correct.
Now, I’m not going to say that there isn’t room for open source here, I could totally see how making, for example, an AI that can look at product description and product appearance and try to classify it for duties would be amazing work if done in open source. I’m just pointing out that the topic is complicated, especially when considering that the same item might have different duties if manufactured in different locations or sold under different circumstances.
Governments are unbelievably complicated even as their systems still don’t mirror or match the real world correctly.