I see what you’re getting at, viz the inconsistency around ‘aided’. However, as it seems that an ‘unaided’ mind will behave anything like ideally I don’t think I’d agree that the inconsistency exists in fact. Isn’t the ‘aid’ here a software package built up by a culture in light of the systematising ideas, and which gets sold on the basis of the mountain of evidence it explains?
pcm50
Karma: 42
A practice I have found useful is estimating the time it takes to do things: length of a journey door to door, time to write a piece of code and get it reviewed, time from setting off shopping to everything being stored away. People are generally on the optimistic side, and time falls through the cracks between stages of a task.
Nice article. I perhaps take issue with this:
And to a large extent, people do pursue those frontiers. It’s no secret that an academic can easily find fertile fields by working with someone in a different department. “Interdisciplinary” work has a reputation for being unusually high-yield.
Since (top tier, excluding Nature, Science) journals typically care most about single axis achievements.
Good call.
I’m a little concerned that this could lead to operationalization of the fundamental attribution error. If you’re in a conversation with someone who says intelligent things, does it matter if they are an intelligent person?
I was just being information greedy. I don’t even know what aspects of health have useful supplements available—so I’m not sure what’s on the menu as it were.
Any suggestions to help (dry) skin conditions would be particularly warmly received though.
The efficient markets hypothesis is that one should expect ‘no remaining region’ to be the default. While betting may not be as competed as finance, there are still hedge funds etc doing betting.
Also I would suggest thinking about expected utility greater than some positive threshold to take into account transaction costs. I suppose that this would make a good deal of difference to how many such regions you could expect there to be.
This is a good idea.
Though I think that the condition that ‘nobody could memorize more than a fraction of it’ is actually quite hard to meet. E.g. legal training seems analogous, and lawyers seem to be able to remember a lot of examples.
If the corpus could be kept secret or ever changing that might help.
When I was thinking of something similar, I had a concern about the task length. E.g. will this result only in relatively short or simple tasks?
You lucked out in terms of your journey, the shortcut through SSC may have saved you a number of years ;)
My only advice on careers would be to strongly consider doing some ML capabilities work rather than pure AI risk. This will make it much easier to get enough qualifications and experience to get to work on risk later on. The risk field is so much smaller (and is less well received in academia) that setting yourself that goal may be too much of a stretch. You can always try to pick thesis topics which are as close to the intersection of risk and capabilities as possible.
For me ‘deeply’ involves coding. I must have read dozens of papers carefully and then failed to implement the technique first time due to either my misunderstanding something, or finding something in the paper to be unclear, wrong or omitted! Or you realise that the technique works well but would scale horribly / can’t be extended etc. That could take several days—so that treatment is reserved for special selections.
I get the point, and it’s a good one. We should have tolerance for the kooky ideas of those who brought us great things.
Practically though, I expect the oracle box would gather dust in the corner if it had a 1% hit rate and the experiments it required were expensive and it didn’t come with an impressive warranty. The world abounds with cranks and it is all too easy for out of the box to get binned with unhinged.
So I think this is a useful mode of thinking with some hindsight on a track record, but not without it.
Me too. I found it strongly reminiscent of reading low grade click bait. Or trying to listen to some woo. Part of it feels like rescuing some food when the bottom of the pan is burned. Part of it is like throwing out models of what state the author’s head was in that resolve the text into sense.
I think what makes GPT2 look relatively good is how low the baseline is in many respects. If you tell me ‘this is a political acceptance speech’ I don’t actually expect it to make that much sense. Most of the genre seems to be written by autocomplete anyway.
Linear Algebra Done Right wins the prize for books which deliver on their title, as promised imho. The typical approach with determinants front and centre is a pedagogical dead end. I find Axler’s approach tracked more closely to the uses of linear algebra, rather than historic proof techniques. I’d say it covers the how and why sections—although perhaps the ‘how’ book is the Matrix Cookbook ;). The what will probably have to come from some other discipline: are you working on stats, or differential equations, or something else?
Arrow on map not pointing to Holborn, suggest checking.
given the rational alternatives (neurology,psychology) we can employ to discover true concepts about morality.
I’m with you most of the way. On the rational alternatives though, I’m not sure what you suggest works in the way we might imagine.
Neurology and psychology can provide a factual/ontological description of how humans manifest morality. They don’t give a description of what morality should be.
There’s a deontological kernel to morality, it’s about what we think people should do, not what they do do.
Psychology etc. can give great insights into choosing morals that go with the human grain. But those choices are primarily motivated by pragmatism rather than vitue. The virtue you’ve chosen is to be pragmatic…
Happy to be proven wrong here, but in terms of what virtues we place value on, I think there’s going to be an element of arbitrariness in their choice.
Varian’s Intermediate Microeconomics is good. I haven’t read Microeconomic Analysis which is apparently more advanced.
The reason I found Varian useful is that principle is paired with math; fairly rigorously up until the later chapters. I also read Mankiw’s Principles of Economics, but you don’t finish that book able to do any economic number crunching.
On the math side it’s mainly differential calculus and algebra. If you have a base there already I suspect it will actually be easier and quicker to absorb the principles from a text with math—since they derive from it.
Sounds good. I hope to make it.
This is an area where technology has made the original method obsolete.
In Obsidian (as I use it) one can add #tags, which are automatically indexed, and [[internal_doc_links]] which can be automatically updated on move / rename.
There’s no longer any need to spend time thinking about how to organize this unique id system as though you needed to put a physical card into a linearly ordered physical pile.
Not doing so reduces the friction of taking notes. For me this is very powerful.