But on the current margin, both small drones and defenses against such seem woefully under-invested-in by Western militaries. Like, several orders of magnitude less spending than would be optimal. It's embarrassing that Ukraine is producing millions of drones a year, and the USA is producing... thousands? Tens of thousands?
I would expect advanced drone programs of the US military are heavily classified. What makes you believe that the numbers you have access to show the true investment the US military is making in it?
If I understand your thesis right it's that you believe your simple math is more likely to result in diagnosing the end of the sleep phase than the machine learning algorithms that are trained-based on the sleep lab data to extrapolate sensor data about heart rate, temperature, movement and oxygen saturation with the correct end of the sleep phase.
Generally the first sleep cycle is measured* to last between 70 and 100 minutes, and the other cycles between 90 and 110 minutes. Those ranges mean there’s some uncertainty in when each cycle actually completes for you; it varies from person to person.
It doesn't just vary form person to person. It varies from sleep cycle to sleep cycle.
To the extend that you do notice improved sleep from what you are doing it's likely because your ritual results in you have a clear intention about when to wake up and your body organizing to meet that intention and not because your beliefs about sleep phases match your actual sleep phases.
Gordon did write https://www.lesswrong.com/posts/kJoyRgDDPzMg4Fo3Z/nice-clothes-are-good-actually for more details.
It's quite easy to buy a air quality monitor that tells you about CO2 or CO but are there monitors that actually tell you about the 100 different substances that might be a problem on airplanes that you can easy have in your carry-on?
I did drop the sequences into alphafold, and I don't see any large structural variation from the SNPs, but (a) that histidine substitution would most likely change binding rather than structure in isolation, and (b) this is exactly the sort of case where I don't trust alphafold much, because "this is one substitution away from a standard sequence, I'll just output the structure of that standard sequence" is exactly the sort of heuristic I'd expect a net to over-rely upon.
Even if the structure is correct and does look the same, the binding properties of the receptor could still be different if the histidine is in the part that's relevant for the receptor binding.
The thing you want is a tool that tells you how the receptor binding properties change through the mutation not the AlphaFold that just gives you the 3D structure. A quick question at GPT-5, suggests that there are freely available tools that tell you how the receptor binding properties change via a single point mutation.
GPT-5 had a botched rollout and was only an incremental improvement over o3, o3-Pro and other existing OpenAI models, but was very much on trend and a very large improvement over the original GPT-4.
I think that's underselling the issue. GPT-5 is not a straight improvement over o3, o3-Pro and other existing OpenAI models. There are tasks it does better and other tasks it does worse.
https://ai.plainenglish.io/i-tried-my-best-to-like-gpt-5-i-just-cant-it-fucking-sucks-7133a1dddfcb shows that even on a task like SQL query generation that you would expect to be easily trainable with synthethic data, GPT-5 performs worse.
I started out giving GPT-5 the doubt but when when it comes to tasks that require some creativity like text and headline writing I went back to o3 as GPT-5 seems worse for the task.
Tulpa have a self-identity, they are not just memes that are passed around.
In his recent All-In talk Elon Musk talked about how he expects all energy by the sun to be harvested sooner or later by AI. This means building a dyson sphere around the sun. The economic reasoning that forces the building of a dyson sphere is quite straightforward.
While a dyson sphere also complete solves our global warming problem on earth, it also has a good potential to create another extinction event on earth.
I wonder whether this might be a scenario that's more useful to talk about when trying to explain the risk of AGI killing humanity.
Does not apply to geneticists and good surgeons in good surgical units. Does apply to many EAs.
What makes you think that good surgeons are good at statistics? I would assume surgeons selects for hands-on-skills. Besides dentists, surgeons are also one of the groups between the medical establishment that care the least about evidence based medicine and practice a lot of treatments where they think placebo-blinded trials aren't really feasible.
Vitamins? physiotherapy? polio vaccine? insulin? CBT? oxygen? abscess treatment? tooth filling? glasses?
Clinical trials of broad spectrum vitamins often find that those don't provide benefits to the broad populations. If we grant that some vitamins do provide benefits, that makes the hypothesis that others in the broad spectrum vitamins have negative side effects quite plausible.
There were times I was getting diarrhea from vitamin effervescent tablets and understanding this made me cut back on them.
Physiotherapy comes in a lot of different forms but plenty of those have side effects. Schroth (which is an evidence-based treatment I got for scoliosis) for example has the idea that muscle tonus should be raised to press the spine into shape which has the side effect of making relaxation with low muscle tonus a lot harder.
Official numbers for the polio vaccine suggest mild side effects in 10-20% of the people that take it for the inactivated polio vaccine we take in developed countries. Risk-benefit calculations suggest that the vaccine is worth it, but pretending that it doesn't have side effects is not helpful.
When asked about side effects of CBT ChatGPT says "Meta-analyses and systematic reviews have documented these “adverse effects” in a non-trivial portion of patients—anywhere between 5–20% depending on the study design. They’re usually temporary, but they matter because they shape whether therapy feels supportive or overwhelming." Generally, a lot of therapy outcomes also depend a lot on the therapist. Depending on the issues that your particular therapist has, you might get related side effects.
Abscess treatment comes in multiple forms. Antibiotics have well known side effects of messing up a lot of other bacteria with whom we live in symbiosis. Cutting comes with scarring risk.
With tooth filling opinions of different dentists about which teeth should get them vary widely. Doing tooth filling quite obviously inhibits the normal healing processes of the tooth and plenty of times tooth fillings break after some time.
Conversely, a medical specialist can be curious about home environment, social life, etc.
A person who bills by the hour (which many alternative medicine practitioners do) has a lot more room to pursue their curiosity in situations like that, then the models with which doctors are usually paid.
I don't think that summarizes best practice. I think it's an important step to ask the AI to ask you to clarify the argument and needs of your writing. You don't want a one-way conversation.