SMBC about Eliezer
Yet another AI alignment problem. Source (https://www.smbc-comics.com/comic/ai-6)
Yet another AI alignment problem. Source (https://www.smbc-comics.com/comic/ai-6)
Monday, March 1, 2021 at 4 pm: https://videocast.nih.gov/watch=41566 (it will be archived at the same link) I thought this would be of interest to people here, especially to those interested in AI alignment. The full announcement follows: National Library of Medicine Lecture on Science, Technology & Society – “Atlas of...
I cannot get the LLM to follow the structure properly. It not only messes up the formatting ~50% of the time, ...
The OpenAI API has a structured output feature that would let you constrain the responses. This will fix the formatting (as long as you have a second phase to transform from JSON to Anki).
it also tends to create cards that are way too long. Splitting them often results in loss of semantic information.
Once you have the JSON, use standard programming/light NLP to check for "too long" and resubmit with instructions that "this <example>" is too long, accumulating feedback in the prompt until you get a set of cards that are short... (read more)
I was frustrated by the lack of a yearly donation option or an option to make a recurring donation of less than $10/month. I almost decided not to give because this communicated that it's not worth the effort to receive contributions from small-value donors like me. And if it's not worth the effort to accept, it's certainly not worth the cost of giving.
However, I decided to give a one-time payment of $10 assuming that this was from ignorance or carelessness. If you'd like to signal that a recurring donation (or more donations from small players like me) are worthwhile, please set up with Liberapay (the GOAT for flexibility—I support plantuml at $0.13... (read more)
Is it a matter of adding a standard paragraph to NIH grants? Yes. That's what I was thinking of.
If you follow standard DEI criteria, I'm commenting on LessWrong; I don't do "standard."😉
More seriously, I apologize. I should have clarified what I meant by diversity. In particular, I mean that diverse groups are spread out in a parsimonious description space.
A pretty detailed example
As a concrete example of one understanding that would match my idea of diversity, consider some very high-dimensional space representing available people who can also do the work measured on as many axes as you can use to characterize them (characteristics of mind, body, experiences, etc.) and reduced by a technique to cause the remaining dimensions to give little mutual information about one another. Define a "diversity-growing procedure" for adding... (read more)
[I] suspect [vaccines] (or antibiotics) account for the majority of the value provided by the medical system
Though I agree that vaccines and antibiotics are extraordinarily beneficial and cost-effective interventions, I suspect you're missing essential value fountains in our medical system. Two that come to mind are surgery and emergency medicine.
I've spoken to several surgeons about their work, and they all said that one of the great things about their job is seeing the immediate and obvious benefits to patients. (Of course, surgery wouldn't be nearly as effective without antibiotics, so potentially, this smuggles something in.)
Emergency medicine also provides a lot of benefits. Someone was going to die from bleeding, and we sewed... (read more)
I shy away from fuzzy logic because I used it as a formalism to justify my religious beliefs. (In particular, "Possibilistic Logic" allowed me to appear honest to myself—and I'm not sure how much of it was self-deception and how much was just being wrong.)
The critical moment in my deconversion came when I realized that if I was looking for truth, I should reason according to the probabilities of the statements I was evaluating. Thirty minutes later, I had gone from a convinced Christian speaking to others, leading in my local church, and basing my life and career on my beliefs to an atheist who was primarily uncertain about atheism because of self-distrust.
Grounding my beliefs in falsifiable statements and probabilistic-ish models has been a beneficial discipline that forces me to recognize my limits and helps predict the outcomes of my actions. I don't know if I could do the same with fuzzy logic and "reasoning by model."
The next post is Secular interpretations of core perennialist claims. Zhukeepa should edit the main text to explicitly link to it rather than just mentioning that it exists. (Or people could upvote this comment so it's at the top. I don't object to more good karma.)
I think you're missing a few parts. The Autofac (as specified) cannot reproduce the chips and circuit boards required for the AI, the cameras' lenses and sensors, or the robot's sensors and motor controllers. I don't think this is an insurmountable hurdle: a low-tech (not cutting-edge) set of chips and discrete components would serve well enough for a stationary computer. Similarly, high-res sensors are not required. (Take it slow and replace physical resolution with temporal resolution and multiple samples.)
Second, the reproduced Autofacs should be built on movable platforms so different groups can get their own. (Someone comes with a truck and a few forklifts, lifts the platform onto the truck, and drives the Autofac to the new location.)
Monday, March 1, 2021 at 4 pm: https://videocast.nih.gov/watch=41566 (it will be archived at the same link)
I thought this would be of interest to people here, especially to those interested in AI alignment.
The full announcement follows:
National Library of Medicine Lecture on Science, Technology & Society – “Atlas of AI: Mapping the social and economic forces behind AI”
Please join us on Monday, March 1 at 4 pm, when Dr. Kate Crawford will deliver the inaugural annual National Library of Medicine (NLM) Lecture on Science, Technology, and Society, entitled, “Atlas of AI: Mapping the social and economic forces behind AI.” The lecture will be videocast and is open to all https://videocast.nih.gov/watch=41566.
Dr. Crawford is the inaugural... (read 403 more words →)
I'm a long-time SR user (over 20 years - I started with a manual system of physical flashcards and a notebook). My rate of card creation is very slow to avoid bogging down my repetitions. I add only a few cards (<5) a day, so I think that "cards from a PDF" or web page would not be helpful.
Tools that would be helpful:
I should also add that I don't use Anki. The UI was awful five years ago (and that's coming from a former SuperMemo user), and it's hard to integrate into my daily workflow. Instead, I use Logseq.