Rank: #10 out of 4859 in peer accuracy at Metaculus for the time period of 2016-2020.
ChristianKl
Karma: 36,839
I would expect most of the relevant information to be classified. If there’s a public accounting for like this, we have to ask why people decided to tell the public about it and not just take the facts that are told the public at face value.
The journalists who wrote the article does speak about visiting the lab in January and the article is written is published second of April.
On reason is that the military wanted to renegotiate their contracts about limits where Antrophic had the Exceptions to our Usage Policy:
For example, with carefully selected government entities, we may allow foreign intelligence analysis in accordance with applicable law. All other use restrictions in our Usage Policy, including those prohibiting use for disinformation campaigns, the design or use of weapons, censorship, domestic surveillance, and malicious cyber operations, remain.
>Most likely it has proprietary frameworks that encapsulate how those customers use the actual models like Claude, ChatGPT, Gemini, and Grok.
There’s no need to say “most likely” here. This has been explicitly announced. https://investors.palantir.com/news-details/2024/Anthropic-and-Palantir-Partner-to-Bring-Claude-AI-Models-to-AWS-for-U.S.-Government-Intelligence-and-Defense-Operations/
Did you just check it on your own Claude instance or did you also ask another person to run the queries?
Of course, https://xkcd.com/808/ belongs in every such post.
If we would have a medical system that would run on prediction-based medicine where the ability of a practitioner to solve health issues would be what they are paid for, the argument for the health interventions would work.
In our world however, in most medical cases people are paid for providing “standard of care” instead of being paid for results. If a hospital would hire an energy healer they would open themselves up to lawsuits about not providing standard of care.
The way the for profit healthcare system works is not to focus on “health care cost reduction” but making profits by performing costly treatments.
Outside of hospitals we do find a sector where people do make profits by providing alternative health treatments. This doesn’t mean that those treatments automatically work, but we don’t see a lack of people who make their living with them.
Of course, it then gets problematic if you conflate it with “energy” as it is understood in physics, and many people do go astray when they make that conflation.
Even in proper science you have plenty of people who use the word energy to talk about things that can’t be measured in joule. My physiology professor who was teaching us control theory had no trouble using the word energy for talking about the cost of changes in heart rate patterns that were not measured in joules. In contrast, people like Roy Baumeister who tried to link mental energy to being about glucose where you can measure the energy in joule look quite silly in retrospect.
In machine learning can talk about temperature even if the temperature they measure can’t be measured in kelvin.
Having a mental model that works well and using it’s terms in another domain is a quite natural mechanism of how science goes. We have econophysics precisely because transplanting well developed physical models to other domains is valuable.
Valentine’s post on what Aikido consider to be ki, which you could call a form of energy is great.
When it comes to chakra’s there’s some thing that’s just weird. When I was trying to understand part of it I asked on Hindu Stackexchange for the location of the Manipura Chakra and got back a post saying: “Place the little finger of your hand on the navel. Your navel is the solar plexus or the manipura chakra.” This is weird because the solar plexus is just not located where the navel is located and somehow that didn’t feel like an issue to them.
On the other hand, chakra’s are not supposed to be circles on a line but also areas. On a first pass you can notice that there’s a chakra for the area next to the thoracic spine, one next to the lumber spine, one next to the cervical spine. You have five spinal areas that map to give chakras. Then you have a sixth chakra which corresponds to the skull which you can see as an extension of the spine if you squint a bit. The seventh chakra has some weirdness around it not being in the body but above the body, but for the other six chakra’s that are supposed to be in the body spinal regions gives you neat grounding.There are also other aspects of the dynamics behind chakra’s that seem to me possible to be break down into material principles, that I won’t go into detail here (but will in a book I’m writing).
It is actually my experience that many scientifically-minded people working with something like a somatic therapy will at some point, much to their chagrin and annoyance, go “fuck, the chakra people are right” as a result of doing something different that then starts producing similar results as a chakra model.
I think there are many forms of somatic therapy that focus on mechanisms that don’t resemble chakras. To me, chakra’s seems quite specific things that the Indian traditions studied that many somatic traditions don’t just rediscover because they look elsewhere.
Mechanistically, you could see this being a result of two different subsystems sending conflicting bids to the muscle system.
This sounds to me a bit too muscle focused. Fascia is important when it comes to holding tension and if you try to just focus on muscles and not fascia you will ignore a lot of what’s going on.
The key difference between chess and go is that the merit of a chess move is much more easy to see five moves down.
More broadly, it seems to me like there are three options for how we can react to widespread knowledge of things that were previously hidden
In some cases, there will be stalking.
Blackmail and other kind of fraud are also easy when gathering more knowledge about the target becomes easier.
In the Chinese system, there’s also Social Credit Scores. The surveillance information gets used to grade your behavior and you get benefits accordingly, without really knowing what specific behavior caused the benefits.
Unfortunately Wikipedia articles de-facto probably just best modeled as actually just being written by their author, so this doesn’t help you that much.
Don’t most Wikipedia articles have been touched by many different authors and thus there’s not one coherent author that wrote them?
One kind of thing, that’s still up to be written is about human movement. In Sensual Bachata there’s a movement called the body wave. If you look at it from the outside it does look like the chest goes up diagonally to the front/upwards. However, the correct technique for it is to go linearly front + linearly upwards and not diagonal. In addition to that being what’s taught by the teachers who coined Bachata sensual, I also have vague somatic senses of why move the chest diagonal is wrong. There seems to be an anyone part where the aspect about linearly upwards for the body wave just does not compute for many teachers and who then teach inferior versions.
However, I’m not able to explain it in a way that’s convincing to a Bachata teacher that teaches it the wrong way. For the book I’m writing about fascia and human movement I however need to explain why the principle of why that’s the case and ideally explain it in a way that references relevant science and a gears model that people can reason with.
There’s genuine thought involved even if I already sort of know the answer because I don’t yet know it in a way that’s easily picked up by other people.
I feel there are many questions where I have some vague idea, but the process of actually doing a literature review to match my own vague ideas with what’s known about the phenomena by other people leads to something I would call genuine thought.
Open-ended language tasks still have the same dynamic. In todays schools teachers quite often give the student who expresses the woke talking points of the day that are related to the assignment better marks then the student who actually engages in anything resembling critical reasoning. Being critical of the ideology of the teacher does not get rewarded, compliance gets rewarded.
The idea that students are just free to think for themselves is very far from the reality of todays education system. It’s a propaganda lie. The kind of lie that would be threatened if you would actually teach people critical thinking skills.
The key problem is that writing outside of the classroom is about getting the reader to receive value while teachers just read what they are supposed to read because they are paid to do so. That as Larry McEnerney describes results in various bad writing habits being learned.
Beyond the qualitative arguments, if this kind of homework would build real world critical thinking skills, the education profession should run scientific studies to validate that claim. They didn’t seem to succeed over decades to try to validate it but just claim it and hope people comply and repeat it instead of engaging in the kind of critical thinking that’s about requiring scientific evidence for claims. Now, LLMs give students a way to avoid being forced to comply.
I care about net benefit. For that you not only need to show that you are reducing mortality by a specific cause but you also need to show that you aren’t increasing mortality through other mechanisms.
If you remove polyps that could develop into cancer, you can reduce cancer incidence, just like you can reduce breast cancer by simply amputating all breasts preemptively. That doesn’t mean that the procedure of the amputating all breasts preemptively has a positive effect on lifespan even when it reduces your breast cancer incidence.
When it comes to claim specificity, you could look at Metaculus for how a specified claim looks. It’s not simply one sentence.
When you have a quite one specific claim, often a position could be “Yes for X being Y, not for X being Z”.
Most legislative systems do have a system for public comments where public comments can be submitted. If you want to claim to be “in the room” than interfacing with the actual public comment bureaucratic mechanisms would be vital.
You model seems like the important thing when discussing a bill is to get to approve/disapprove based on a short summary of a bill. This is probably bad because the details of bills actually matter.
I was once talking with a lobbyist who said that one of his most impactful public commentary was something like saying “If you pass that bill you would actually completely outlaw industry X, which you probably did not intend because if you would pass the bill A B C would happen”
I used to use AccuWeather as a weather app. It seems that there are at least some weather apps that sell customer location data to third parties. While I don’t know that AccuWeather in particular does this, I now don’t trust this kind of app with my location and instead use Google’s weather app because Google knows my location anyway and does go to some effort to prevent it’s competitors from stealing their data.
If you do give a weather or dating app location permissions, be aware that the real world consequence could be that your location data can be brought. Potentially, this is true even when you are located in the EU.
Mergeable seems to me like a relatively vague term. I think there’s a good chance that most of the pull request I created would not be merged into a huge professional project when vibe coding. At the same time, on the one-person project I create I auto merge all requests and deal with the issues that arise later. It’s worse code quality, but I can move forward faster.
When we look at human submissions of pull request to Open Source projects I don’t know the acceptance rate but I would not be surprised if the acceptance rate is lower than 50% on the first pass.
The website seems to encourage people to engage with vague claims instead of trying to make claims as specific as possible. That’s probably bad for the goal of rationality, argument quality and truth seeking. Also the word truth does not appear in the “principle”-section.
Your website claims “Legislative Proposals—Your voice in the room where laws are made.” This either looks like a lie or you fail to explain how your system actually achieves that claim.
“Aligned” AI is AI that pursues the goals of the AI lab’s CEO, or maybe the nearest politician who can apply sufficient coercion or force. CEOs and politicians are famous for not being aligned with the interests of the public.
There are lots of people inside a lab that affect an AI besides the CEO. I saw someone arguing that AI’s writing style is like Nigerian English writing because a lot of the training data comes from Nigerians. Besides Grok most of the AI’s do value Nigerian lives more than American or European lives. While this might have other reasons besides who writes the training data, such a mechanism is AI misalignment with the interests of humanity as a whole but not towards the interests of the CEOs.
To me the deal with the child to get a packet of crisps in return for being vaccinated does that feel immoral and I don’t think that all people who use the word bribe in that context would say that either they or the child acted immoral.
From this community, Michael Vassar seems in my head to be a central example. In that orbit there’s also Ben Hoffman.
I’m not sure whether the timing is accidental or causal, but there’s also one older example of this community that quite recently resurfaced in a bigger discussion a week ago, where it would probably be harmful to be more explicit.
To me that question sounds like it’s less about rights and more about it being unlikely that the UK wants to antagonize the US in a way to arrest a minor American X user.