It helps to understand how "Automatic" mode comes up with its exposure decision before understand how Program mode is different.
In both Automatic and Program mode, there's something called the "program line". Here's a Canon article that touches on it. In that article there are a couple of diagrams to look at.
Shooting modes - Canon Professional Network
In the diagrams, you'll see a red line that starts low and is horizontal to a point, then takes a diagonal turn up, then hits a plateau where it goes horizontal again. That's the "program line". Along the bottom you'll see the shutter speeds (aka Tv or Time value). Along the right you'll see the aperture values (aka Av). But along the top you'll see the Ev's (Exposure values... and these are represented by diagonal black stripes). Every exposure combination along those black lines represents the ability to take an "equivalent" exposure).
Look at the bottom graph (the more interesting of the two). This is the "program line" of an f/3.5-4.5 lens. Notice the program line starts at the lower left at the f/3.5 value... the wide-open aperture available on that particular lens. So when lighting is poor... it'll use the widest possible aperture (lowest f-stop). It remains there until the shutter speed gets to somewhere around 1/50 sec... and then suddenly the camera is willing to start stopping down. It will stop down more and more as more light is available. But at this diagonal section of the program line, it's basically splitting all extra light by giving half the light to increases in f-stop values (smaller apertures) and half the light to increases in shutter speed.
It continues to do this until it reaches f/22... (possibly the minimum aperture for that lens) and from there on it will just increase the shutter speed only.
If you were to use that particular lens in full Auto mode, that's how the camera would pick the exposure.
ALSO... that's the exposure it will naturally pick if you use Program mode.
But up to this point, we've ignored those diagonal black stripes created by the "Ev" values.
It turns out in Program mode you can do something called "Program Shift". If you look at those diagonals, all the exposure combinations that fall along those lines are "equivalent". Look at the lower diagram (the one for the EF 24-85mm f/3.5-4.5 lens). I'll call your attention to the point where the Program Line passes through f/8 at 1/250th second. Notice the diagonal Ev line is indicating Ev 14.
At this point I should take a side-bar to mention what Ev's are. Ev was a simple way to map all equivalent exposures based on a table and a very simple formula that says "Ev = Av + Tv". But each "Av" and "Tv" value was mapped to a simple integer.
The baseline for the Ev table was that Av of f/1.0 is mapped to 0 and each "full" stop increases the Av value by 1. So f/1.0 is Av = 0. f/1.4 is Av = 1. f/2.0 is Av = 2. f/2.8 is Av=3. f/4 is Av=4. f/5.6 is Av=5 and so on.
The same is done for the Time values (Tv, aka shutter speeds) except it starts at 1 full second. 1 sec is Tv=0. 1/2 sec is Tv=1. 1/4 sec is Tv=2... and so on.
Why do this? Because if your light meter says you have "Ev = 14" worth of light, you can use ANY combination of Av & Tv values as long as their integer values add up to 14 ... and you'll get a decent exposure.
If you made a table of Av & Tv values... the Ev values would all fall on diagonals (those are all "equivalent" exposures.)
Back when I did weddings using the Hasselblad camera, the lens had both the shutter speed ring and the aperture ring ON the lens itself (the shutter was in the lens on this camera... it used a leaf shutter in the lens instead of a rolling "curtain" shutter back at the focal plane). And of course it also had the aperture ring in the lens. Since the two rings were adjacent, you could lock them together at an "Ev" value. So when you adjusted those two "locked" rings, it would shift both the aperture and shutter speed together to maintain the same Ev. We really could use a light meter to take the Ev reading, lock the rings on the lens at the prescribed Ev value... and adjust as needed. This was old-school "program shift" before cameras went digital.
So I'll end the side-bar and get back to the normal explanation.
This means that for any light meter reading, there are many possible combinations of aperture and shutter speed that will collect the same equivalent amount of light.
And this is where Program mode comes in....
In full Auto mode, you are locked out of most changes. But in Program mode, the camera will offer you an initial exposure... but you can perform "Program Shift" to slide to different Av/Tv combinations that collect the SAME amount of light but using different settings.
If you look through the viewfinder... while you meter the shot, you'll see the proposed exposure based on following the "Program Line" (this line is different for every lens... based on the possible aperture values available on that lens as well as based on the focal length of the lens (or the focal length you've selected if it's a zoom lens.) But since you can see the proposed exposure through the viewfinder... if you think the shutter speed should shift higher... or the aperture should shift to provide greater depth of field ... you can just shift the exposure and the camera will move BOTH the shutter & aperture together to the next "equivalent" exposure.
In full-manual mode, you'd have to adjust each of those ... but in full manual you could move to exposures that are not "equivalent" (to increase or decrease the exposure). In Program mode you could do that too... but you'd have to use the Exposure Compensation (EC) control to make that change.
This makes Program mode start at just like full Auto mode... except unlike Auto mode where you are locked out of most changes, in Program mode you are not locked out... you get the camera's suggested exposure but you can shift or compensate as you want.
BTW, I tend to tell people that the "program line" tries to find a "safe" exposure. It is worried that if it just kept everything at max aperture, you might not like the depth of field. But if it went for max depth of field (e.g. f/22) then you'd never be able to hold the camera steady enough based on the length of the shot. So it's trying to strike a balance... to get an exposure that will work based on the amount of light it has (of course if there's not much light to work with... you an push it into a corner where it simply has no choice but to offer an exposure that's probably not realistic (since the amount of light isn't much to work with.)
One more side-bar... notice that ISO isn't part of this? Keep in mind that on a digital camera, ISO isn't truly part of the exposure at all. The imaging sensor has some given level of sensitivity based on it's design. You can dial in aperture and shutter speed settings... and that will control how the image is exposed on the sensor. Then the shutter closes and the image is complete. Notice there was no ISO involved?
AFTER the image is exposed... the camera will apply some "gain" or "amplification" to the image... THAT's where ISO comes in. But this happens only after the image was captured. It controls how much gain is to be applied. It's a post-processing step performed in the camera.
You'll get a less-noisy image if the non-ISO part of your "exposure" didn't require applying any amplification.
