It is with great sorrow that I note the passing of Marvin Minsky, one of the great minds in the field of artificial intelligence, at age 88. While his achievements in computing at MIT are well-chronicled elsewhere, such as this New York Times obituary, I’ll remember him as a TidBITS reader and for his openness and kindness. Although plans to meet up at Macworld Boston never came to fruition, I exchanged email with him back in 1992 about Apple’s Casper voice recognition technology (see “Casper Speaks,” 9 March 1992), which evolved into a system called PlainTalk (see “Bossing Your Mac with PlainTalk,” 28 August 2000) and which is still in OS X today, as Scholle McFarland explains in “El Capitan: A Take Control Crash Course.”
In that 1992 conversation, Marvin wrote about his worry in regard to speaker-independent recognition, which at the time seemed like a tall order. (Keep in mind that this is the era of the “wicked fast” Macintosh IIfx, and well before the Internet became popular.)
In fact, I don’t doubt that Apple and Kai-Fu are way ahead, etc. It is only that I have much concern about “speaker-independent” recognition in any case. When a stranger says a few words, you don’t always get it — and I suspect that we usually start with a bit of small talk, not only to establish social ease, but to establish a phoneme lookup table! I hope they don’t kill the golden goose by trying to avoid this! In fact, you could do this just once, by being asked to repeat a sentence or two; then I should think that the system could incorporate a crude “voiceprint” system to recognize who you are, etc. I’d rather talk to my workstation for a few minutes than be constantly correcting errors. Well, of course they’ve thought of all that!
He wasn’t wrong — although there have been speech recognition systems since that time (Matt Neuburg wrote about one, with background on the field, in “Talk Is Cheap — ViaVoice Enhanced Edition,” 21 August 2000), speaker-independent recognition has become truly widespread only in the past few years with the releases of Siri in iOS, Google Now, and Microsoft’s Cortana (Tom’s Guide has a nice comparison of the three voice assistants). And even these technologies are far from perfect.
On a more personal level, Marvin closed that discussion with these kind words.
And thanks for your terrific public service. I read TidBITS frequently and, as a result, my IIfx has more control panels and extensions than is safe. Needs booting pretty often!
I don’t know for how much longer Marvin kept reading TidBITS, but I’ll cherish the thought that our efforts were of use in his work, even if only in an everyday computing environment.