I just ran some tests; my 28 2.8 AIS vs. my 28 3.5 K (AI'd). BTW, I used both lenses on a Canon EOS body (with half-frame sensor) via the same adaptor. Thus, focus depended on my eyes only w/ neither a split prism nor a hot focus point to rely on. On a good Nikon film camera, the results might be a bit different.
Note also that my tests were oriented toward my interests: nature photography. I often shoot into the sun so I always test for flare. I also often use scale focusing stopped down. Notice however that much of my testing was indoors where I can control the light. It's a windy, partly cloudy day, making lens testing a little hard.
So here's my entirely subjective conclusion:
Indoors, close focus: 28 f3.5 K is noticeably sharper at every aperture from 3.5 to 11 (did not go beyond that). Even the blurry areas (beyond DOF) were sharper on the 3.5.
Indoors, lenses focused at their hyperfocal distance: roughly equal in the centers, a slight advantage to the AIS in the corners.
Outdoors, lenses focused at their hyperfocal distance: roughly equal in the centers and corners. Some advantage to the AIS on objects in the infinity distance.
Flare: Advantage to the 3.5 K (AI'd) when including the sun in the frame. Both flare a bit, but I didn't find the 3.5 to be objectionable at all whereas the 2.8 can have some large annoying zombie blobs growing out of the sun.
Notes:
- For those of you that scale focus, the 3.5 is much easier to use than the 2.8. I noticed the hash marks on the chrome are farther apart on my 3.5 than the 2.8. I appreciate that.
- Build quality is excellent on both lenses.
- Both are multicoated, but since the AIS I have may have been manufactured 10-15 years later than the K, I suspect its multicoating might be, I dunno, better?
Lastly, I got my 3.5 AI'd K lens on eBay for $65 + shipping. I believe I got the AIS on KEH for about $200. In terms of bang for the buck, a big advantage for the older lens!