I believe that the best way to think about the problem is to consider light intensity to be a measure of photons/second.
Any light source generates a bunch of them per second, so you use various techniques, like pinholes, partially silvered mirrors and neutral optical density filters to reduce the number/second until discrete events are detected. This all happens in a completely darkened chamber, of course.
You're technically not generating a single photon. The purpose of the experiment is to prove that light is actually a wave.
This is the pattern you'll find when you pass light through a double slit:
This is the theoretical photon probability density:
So the point is to find the number of photons at different maxima and try to prove or disprove this probability density.
If your question was how does the machine generate a pulse for each of those photon? then the answer is that light travels in a "straight line" and the photon detector is so fast and powerful, that it detects each photon individually and sends out a pulse that is only 10ns wide.
No, sorry, that's not my question. I can imagine how to detect individual photons, I cannot imagine how to generate them one at a time. @jremington has given me a pretty good answer but maybe if this is part of your studies you know more. My understanding of the double slit experiment is that even if you send the photons one at a time they still interfere with themselves, so show wave like properties, not particle like properties. I have no difficulty with the concept of the experiment but I cannot think how you could possibly build a light source that generates only one photon at a time, so my question is: how do you construct a light source that emits one photon at a time?
I looked at the Harvard demonstration kit and the description and I am shocked. The basic demonstration is fine, all makes sense.
What does not make sense is the polarising filters that can be inserted to track which slit the electrons go through. If the filters as orthogonal to each other then of course there is no interference pattern, how can there be with polarisation at 90 degrees? This is so obvious I cannot understand why there is any debate about it. Adding the 45 degree filter will restore the pattern because the polarisation is now the same again for both slits.