Maybe it all boils down to copyright. Having a method that believably removes the capacity to generate copyrighted results might give you some advantage with respect to some legislation.
Also if you build some sort of search engine using an LLM governments will expect you to be able to remove websites or knowledge of certain websites for legal reasons (DMCA, right to be forgotten, etc).
How do you get a dog to stay as your friend? You don't have to "make" a dog stay your friend assuming of course you treat it as a friend but also because the inherent domestication has made the dog inclined to remain your friend, such an AI would be inclined to remain your friend.
Modules are a massively over engineered "solution" to the problem that require significant refactoring to actually make use of them. Have you tried to properly use modules (i.e. create ones in your software, not just import std)? It's super clunky and still hardly usable.
I doubt we'll see Unreal Engine get any benefit from that in a long time for example. It could be so much better, working fully automatically with almost all existing code so long as you use IWYU, which is already standard for large projects where this is needed the most.
There is just too much randomness in life and control is an illusion. Of course, it is true that some life choices will close or open doors, but, again, life is messy for everyone.
By the way, when did mindfulness fall out of favor with tech people? I have the feeling there is a whole new generation that again has to learn the benefits of contemplative practices from scratch. Anyone here read Search inside yourself etc. ten years ago?
Offenbach also hosts a relatively large Japan festival (in Germany). Wood joinery, at least in my mind, is a very Japanese craft. I wonder if this is by accident.
It focuses a lot on the evolution of precise woodworking tools, like saws and planes. They also had examples of complex joints, made without nails or glue.
There are a couple of Japanese joinery books out there. One that I've read gives the reasoning for all of the joints in the Japanese tradition for joining two boards end to end (thereby creating a longer board, in effect) is that the timber available for building in Japan doesn't come in long, straight lengths.
I can't tell you if that's true or not, but I have a hard time seeing folks putting so much work into developing such joints without a pressing need for them.
This thought experiment is deeply flawed as it assumes that what you propose is possible and the outcome would be as you suggest. I am not convinced that people can bootstrap themselves like this and that language itself does not contain views, values and perspectives.
Step one with someone locked or frozen-stuck in such belief-logic is that improvement is possible - which is part of developing psychological flexibility.
You'd have to pay attention and find any step forward possible for them to begin to enter the discomfort that is holding them back - which may be the biggest challenge of their lives up until that point.
Ideally though, as we're all sheep to some degree - which is exactly what this HN post is stating in a more sophisticated way, and attempting to warn for this - ideally there's a culture of practices that develop oneself, so you're just going along with the "herd" - that path hopefully not corrupted and led by bad actors attempting to send us off a cliff or into their totalitarian pen.
> that language itself does not contain views, values and perspectives.
Language itself does, in fact, very much contain views, values and perspectives. As an example, there are population groups that do not have a word for the color blue, and that cannot, in consequence, distinguish between green and blue objects. [0] [1] And that's an example that has been noticed throughout the world.
The claims in the first link, at least, should set off your BS detector. The colors they claim the group could trivially identify are very near identical in terms of 3d distance. By contrast, the colors they claim they could not determine the difference between are extremely far in 3d distance. It's objectively illogical. A quick search turns up that the claims are indeed false, and were fabricated by the BBC for a documentary. Here's [1] an email chain involving various researchers that worked on these experiments. The conclusion is that:
---
"The experiment shown in the documentary was a dramatization; the genuine color experiments done with the Himba, some years before, used a different sort of stimuli and a different experimental method; the stimuli shown in the documentary were modeled on those used by Paul Kay and others in experiments on other groups; but in all of the relevant experiments, the dependent measure was reaction time (in finding a matching color or an oddball color), not success or failure.
The BBC's presentation of the mocked-up experiment — purporting to show that the Himba are completely unable to distinguish blue and green shades that seem quite different to us, but can easily distinguish shades of green that seem identical to us — was apparently a journalistic fabrication, created by the documentary's editors after the fact, and was never asserted by the researchers themselves, much less demonstrated experimentally."
I did not do a full literature study of the claims of the article, as that was meant as an illustration of the point, and it was the first relevant link that popped up.
Whatever the alleged fabrications of the BBC might have been or not been for the Himba experiments, similar observations have been made for other groups as well, from Amazonian tribes, to population groups in Papua New Guinea, to Aborigines in Australia. And the MIT research link is discussing a very similar result.
That languages influence how you perceive and see the world has been well studied and is well documented.
I do not think the actual experiments demonstrate this. The experiment the BBC demoed probably would, but it was fabricated. The actual experiments just demonstrate a pretty obvious aspect of learning. If you tell somebody who doesn't play chess all the names of the pieces, and ask him to tell you which is the rook, he's going to be slower and less accurate at it than somebody who plays chess. But there's not some huge epiphany like you can show him a collection of 5 bishops, 1 rook, and him be unable to tell which is different. It's no different with colors, or any new term.
I have personal experience with this, learning Russian. They don't have just blue, but rather a term for dark blue and one for light blue. It's hardly some eye opening thing - it's basically sky blue vs ocean blue. It's obvious and easy, but obviously I will always be slower than a native on a quiz of which is which for reasons that have nothing to do with the colors. Vice versa, compare our speeds in English with 'sky blue' vs 'ocean blue' and I'd be back to winning.
A common trend in the social sciences is creating experiments that aren't designed to challenge one's hypothesis, but confirm it. The publication bias against negative results is probably necessary, but also turning a lot of soft science into a facade.
