Posts

Brief notes on the Senate hearing on AI oversight 2023-05-16T22:29:33.167Z
List of commonly used benchmarks for LLMs 2023-04-20T02:25:01.947Z

Comments

Comment by Diziet on Survey: How Do Elite Chinese Students Feel About the Risks of AI? · 2024-09-05T19:31:43.287Z · LW · GW

I wonder how much the survey paying a few RMB and asking for a wechat ID influenced the results. Great work working and sharing this poll though!

While this survey’s responses are anonymous, respondents did submit their WeChat IDs so that they could be remunerated for participation.

Comment by Diziet on Jaan Tallinn's 2023 Philanthropy Overview · 2024-05-21T00:32:15.013Z · LW · GW

Top 10 donations in 2023, since the html page offers no sorting and is sorted by date:

$2,800,000Cooperative AI FoundationGeneral support
$1,846,000Alignment Research CenterGeneral support for ARC Evals Team
$1,733,000Center for Applied RationalityGeneral support for Lightcone Infrastructure
$1,327,000Center on Long-Term RiskGeneral support
$1,241,000Manifold for CharityGeneral support for Manifold Markets
$1,159,000Alliance to Feed the Earth in DisastersGeneral support
$1,000,000Carnegie Mellon UniversityFoundations of Cooperative AI Lab
$1,000,000Massachusetts Institute of TechGift to the Tegmark research group at MIT for General Support
$1,000,000Meridian PrimeGeneral support
$909,000Center for Artificial Intelligence SafetyGeneral support
Comment by Diziet on Stuart Russell and Melanie Mitchell on Munk Debates · 2023-07-17T20:04:59.858Z · LW · GW

The current link to the podcast is available here: https://munkdebates.com/podcast/the-rise-of-thinking-machines

Comment by Diziet on What if they gave an Industrial Revolution and nobody came? · 2023-05-21T19:53:21.664Z · LW · GW

It seems one could convince this hypothetical emperor to invest into industrialization of technology by offering to build things other than a steam engine, or outlining how a steam engine leads to them -- telegraph, or semaphore towers to send news of invasions or changes in distant towns or provinces, better manufacturing capability for tools and weapons, food storage and transport mechanisms, etc.

Comment by Diziet on Use these three heuristic imperatives to solve alignment · 2023-05-18T00:59:19.652Z · LW · GW

I looked over a bit of David's public facing work, eg: https://www.youtube.com/watch?v=I7hJggz41oU

I think there is a fundamental difference between robust, security minded alignment and tweaking smaller language models to produce output that "looks" correct. It seems David is very optimistic about how easy these problems are to solve.

Comment by Diziet on Brief notes on the Senate hearing on AI oversight · 2023-05-17T17:54:45.079Z · LW · GW

I tracked down the exact quote where Prof Marcus was talking about timelines with regards to jobs. He mentioned 20-100 years (right before the timestamp) and then went on to say: https://youtu.be/TO0J2Yw7usM?t=2438

"In the long run, so called AGI really will replace a large fraction of human jobs. We're not that close to AGI, despite all the media hype and so forth ... in 20 years people will laugh at this ... but when we get to AGI, let's say it is 50 years that is really going to have profound effects on labor..."

Christina Montgomery is explicitly asked "Should we have one" [referring to a new agency] by Senator Lindsey Graham and says "I don't think so" at https://youtu.be/TO0J2Yw7usM?t=4920

Comment by Diziet on [deleted post] 2023-05-10T20:53:18.928Z

Couple of more takeaways I jotted down:

PaLM2 followed closely [to] Chinchilla optimal scaling. No explicit mention of number of parameters, data withheld. Claim performance is generally equivalent to GPT-4. Chain-of-thought reasoning is called out explicitly quite a bit.

Claims of longer context length, but no specific size in the technical report. From the api page: "75+ tokens per second and a context window of 8,000 tokens,"

"The largest model in the PaLM 2 family, PaLM 2-L, is significantly smaller than the largest PaLM model but uses more training compute" "The pre-training corpus is significantly larger than the corpus used to train PaLM [which was 780B tokens]"

Comment by Diziet on TED talk by Eliezer Yudkowsky: Unleashing the Power of Artificial Intelligence · 2023-05-08T00:10:31.882Z · LW · GW

I was somewhat disturbed by the enthusiastic audience applause to dire serious warnings. What are techniques or ways to anchor conversations like this to keep them more serious?

Comment by Diziet on Stupid Questions - April 2023 · 2023-04-08T00:02:51.622Z · LW · GW

As a new user -- is it ok and acceptable to create a new post? I have read the discussions in this community in logged-out-mode for quite some time, but never contributed.

I wanted to make a post titled "10 Questions and Prompts that only an AGI or ASI could answer"