CV is older than the 1960s, and Minsky had enough experience with it to know it wasn't an undergraduate summer student job.<p>Minsky and Russell Kirsch worked on CV back in the 1950s. Quoting <a href="https://conservancy.umn.edu/bitstream/handle/11299/107503/oh179mlm.pdf?sequence=1&isAllowed=y" rel="nofollow">https://conservancy.umn.edu/bitstream/handle/11299/107503/oh...</a> :<p>Minsky: From the very start AI labs were obsessed with making machines that could see. So there was the early vision work. A lot of the early vision work was somehow associated with AI. Although my friend Russell Kirsch at the Bureau of Standards, had been doing it as early as 1954 or '55.<p>NORBERG: Which? Vision?<p>MINSKY: A little bit of computer vision with the SEAC machine. Nobody used the word AI yet. He was the kind of person who could say well, everybody's having their machines crunching numbers, can we have one that can recognize patterns. Of course, the electrical engineers would dispute that and say, we've been making sensory systems all the way back to World War II.<p>In <a href="https://www.webofstories.com/play/marvin.minsky/29" rel="nofollow">https://www.webofstories.com/play/marvin.minsky/29</a> you can hear Minsky talking about recognizing letter shapes in the 1950s. Though I don't know if his work used images, or simply a graph data structures.<p>Minsky's 1960 "STEPS TOWARD ARTIFICIAL INTELLIGENCE" (see <a href="https://web.media.mit.edu/~minsky/papers/steps.html" rel="nofollow">https://web.media.mit.edu/~minsky/papers/steps.html</a> ) also describes some of the work toward training a computer in "recognizing normalized printed or hand-printed characters."