This fascinating app lets one determine the exposure time in various settings (assuming those settings include at least one infected individual) with a given risk tolerance. https://indoor-covid-safety.herokuapp.com/ 

It is based on this paper (I think?), in case you want to read the details of the model used. https://www.medrxiv.org/content/10.1101/2020.08.26.20182824v2.full-text

Here's an example, I think representing a typical classroom setting (perhaps the room itself is larger than a typical classroom, I'm not sure).

According to this app and the settings I've selected, if you have a 25 person classroom in a typical classroom environment, assuming minimal talking occurring, after 52 minutes everyone has a 1% chance of being infected, assuming also that there is 1 infected person in the room.

Those 1% chances of being infected are independent from each other and related to individual risk not collective risk, as far as I can tell, so there's about a 22% chance of someone in the room becoming infected in that time-frame.

For what it's worth, if there is a 3% positivity rate in the community, assuming equal distribution of that rate (this is a false assumption, by the way, but sometimes one needs to approximate), then there is a 53% chance that at least one person in a classroom is infected.

If we want to reduce the risk of infections spreading, then we need to reduce class sizes A LOT.

Here are some pretty important caveats to this modeling.