Having read that, I think the linked article is using "transparent" in a way that the original article never did, which is the cause of so many commenters' confusion.
The article never says the film is more transparent than glass. It says
> The average global transmittance of the PMMM in the sunlight spectrum range is 95%, surpassing that of glass (91%). This is because the incident rays reflected at the micro-pyramid’s surfaces can be redirected to the PMMM
My understanding (someone correct me if I'm wrong) is that this film cuts down on reflected glare, and redirects that light downward, so more light ends up going through the glass.
This makes the glass more translucent, which is the ability to pass through light, but not more transparent which I understand to be the ability to see clearly through glass.
You're right about what it does, but the original article also often calls this material "transparent", e.g.:
> This design integrates several functionalities, including light diffusing, self-cleaning, and radiative cooling, while maintaining a high level of transparency [emphasis mine].
A better quote that I think illustrates that the paper was confused about the terms as well:
> The combination of these features makes PMMM a practical solution for transparent roofs and walls, offering improved light management, energy efficiency, and occupant comfort. Moreover, the use of readily-available, affordable, and environmentally-friendly polymer materials ensures the potential for large-scale manufacturing while remaining competitive with existing transparent roof and wall materials. Overall, the development of this multi-functional metamaterial paves the way for sustainable green buildings with enhanced transparency, energy efficiency, and occupant well-being. It contributes to the ongoing efforts towards creating a more sustainable built environment.
I think it's fair to summarize this as "a roof coated with this material would be more transparent than a transparent roof today", which is wrong. It could be interpreted as saying "by reducing the disadvantages of a fully transparent roof, this will encourage more buildings to adopt roofs that are not fully opaque, thus enhancing the transparency of roofs in buildings overall", but I think that's needlessly confusing.
This paper and other similar works changed my opinion on that quite a bit. It shows that to perform text prediction, LLMs build complex internal models.
You can pass some knowledge but it's limited. The more specialized you get the fewer knowledge bearer and if something happens to them you have to start from scratch again.
Yes there is a day/night cycle but it also depends too much on the weather on a ridiculously tiny scale to be realisticly predictable. And even if it was indeed predictable, it woud still add complexity to the prediction
reply