Relative to its competitors, Anthropic seems to have a higher share of professional users paying premium subscriptions, which is probably more sustainable in the long term.
1,477 karma · joined May 14, 2013
Nevertheless, we don’t have a good conceptual framework for thinking about these things, perhaps because we keep trying to apply human concepts to them.
The way I see it, a LLM crystallises a large (but incomplete and disembodied) slice of human culture, as represented by its training set. The fact that a LLM is able to generate human-sounding language
And from the point of view of the maintainers, it seems a terrible idea to set up rules with the expectation that they will be broken.
Personally, I remain skeptical that is the case.
What does seem likely is that “intelligence” will eventually be redefined to mean whatever we got out of the AI buildup.
One of those use cases is training in industrial settings like factories, mines, or construction sites. It is a lot cheaper and safer to have workers take their first steps in a VR environment rather than on the actual location.
At the moment, a lot of this training is implemented as custom 3D applications using e.g. Unity. I imagine that something like Hyperscape would make it much easier and faster to create and share VR environments for this purpose.
On the one hand, this reflects the US dominant position in world affairs and the fact that probably most of the training materials come from there.
But on the other hand, there are some outlier results that are left unexplained. For example, ChatGPT is even more aligned with Japan than with the US.
European and American maps place the Atlantic in the middle, because it minimises the distortions to those regions and makes them more visible. Asian maps put the Pacific in the middle for the same reason.
Reading the article, I am reminded of the medieval maps that put Jerusalem in the center, with Asia at the top and the Mediterranean flowing down from it. A spiritual map.
Perhaps what the article is describing is also a spiritual map in its own way.
However, that was a result of faulty assumptions made when the plant was initially planned. With better data and methods, the event would have seemed a lot more likely.
I understand that people assume LLMs are private but there isn’t any guarantee that is the case, specially when law enforcement comes knocking.
It is very useful to be able to ask basic questions about the code that I am working on, without having to read through dozens of other source files. It frees up a lot of time to actually get stuff done.
At the same time, more capable models are also a lot more expensive to train.
The key point is that the relationship between all these magnitudes is not linear, so the economics of the whole thing start to look wobbly.
Soon we will probably arrive at a point where these huge training runs must stop, because the performance improvement does not match the huge cost increase, and because the resulting model would be so expensive to run that the market for it would be too small.
As one commenter notes, we seem to be heading towards a “don't ask, don't tell policy”. I do find that unfortunate, because there is great potential in sharing solutions and ideas more broadly among experienced developers.
However, that way of working can be exasperating for those who prefer a more deterministic approach, and who may feel frustrated by the sheer amount of slightly incorrect stuff being generated by the machine.
Unfortunately, people are swallowing the headline without any critical thinking.
Furthermore, kanji often allows the reader to skip phonetics altogether, because the symbol itself carries the meaning. It’s a bit like how people understand emojis without having to make a sound for them in their heads.
As a result, the Japanese are able to read very fast in their native language, so switching to a different system would carry significant drawbacks.