I really, intimately collaborated with two major professors at Stanford, in very different disciplines. That was one of the attractive things about Stanford, the entire campus, everyone is together, the medical school was across the street from us, which is usually not the case, the medical people are very far away from the chemists. And that was Joshua Lederberg, who became a very close friend of mine, who at that time was the Chair of Genetics, but he’s an absolute genius, with interests in many other areas. And the other one was Ed Feigenbaum, who was at that time the Chair of Computer Sciences. And Ed Feigenbaum was one of the pioneers of artificial intelligence. Joshua Lederberg, who knows everything, was among other things very much interested in exobiology, namely life in outer space, and was involved in some of the preliminary work on some of the NASA missions at the time, and is still focused on unmanned exploratory things, and therefore wanted to develop methods of detecting natural products, if they existed, on Mars and on the Moon, by methods which had all have to be spectroscopic, and Feigenbaum... Well, he and Feigenbaum got together and realised that there was a lot of computer stuff that has to get involved in this, because then it’s only through telemetry and computer interpretation that you can go and analyse these data, and then became interested, and it was just the beginning of artificial intelligence, and as I said, he, Ed Feigenbaum was one of the pioneers in this. And he first started collaborating with Joshua Lederberg about this, and a couple of years later approached me and said, how about using it in chemistry, and we’ll use you as an example, and say, we’ll let the computer learn how Carl Djerassi thinks, and let’s see whether you can put that into a program language, and that led to a series of papers. We probably published something in the order of 40 papers on this, which had always the title, Application of Artificial Intelligence to Organic Chemistry, and that became very interesting, so we did an awful lot of work on this as well, and computerised even the structural elucidation of natural products, which are then used very much pedagogically, and I’ll explain this in a very amusing experiment... illustration.

The manner in which an organic chemist thinks, even to this day, when he doesn’t do any chemistry any more, but only uses spectroscopic methods, when he wants to arrive at the chemical structure of a compound, and say, you know, this is a structure of a compound, this here, and you can see that this has nothing to do with steroids, it looks very different. Or this, the chemical structure, well, you know, you don’t see it, you don’t have it in your hands, you, you infer it, and what you’re doing is, you draw up straw men, alternative ones that are consistent with the data, and then you try and find a piece of datum... I always... for me data was the plural... a piece of datum that eliminates one but not the other. And eventually what you do is you kill all the other straw men and you’re left with only one, but he’s still a straw man in a way, although emotionally you are convinced logically that really exists, but you don’t really know that. You will only know this under two conditions, either if you actually then make it unambiguously, and therefore, then you’ve confirmed it, because then you have it in your hands, so to speak, or you take a flash picture, an X-ray picture, which shows you there it is, even though you don’t have it, but there it is. Well, the computer, while it can think completely by himself, and I purposely always visualise him as a man, as a dull man, and not really as a sexy woman, is... the computer can operate much faster than you can, although with the fundamental things it may be a different proposition. So the computer can analyse all the data that you have, and compare them with the straw man that you have drawn up. It can do more than that, it can create straw men, all the straw men that are consistent with the data that you have, and it turns out that almost always the computer can think of more straw men than you can, which is something which is humbling and also quite interesting. Now, we... when you do that, and so he comes with some more ideas that you didn’t have, but then he draws pictures, theoretical ones, and then, of course I the intelligent chemist can say, ah, but I could devise a method that will separate that new straw man from the other one, and eventually you still reduce it to only one, but by having the confidence that the computer came up with all possible straw men, by reducing it to only one, then I think you are there. So I used that in an advanced course in chemistry with graduate students, where basically we taught them how to use that software program, and then said, you go into the literature, each of you pick on your own, a different paper that’s been published by different people, which purports to describe the chemical structure of a new natural product, and sometimes it’s quite complicated. Where they then publish it and say, this is it, proudly, and they’ll take all these data, which are described in there, and put them in the program, and tell the computer, can you think of any other straw men? Well, it turns out, in every paper the computer thought of at least one more, but in some cases as many as 40 more, and then I said, now write to these authors, who happen to be in many different countries, and say, dear Doctor So and So, you published this paper that the structure of compound X, Y, Z, is this, but I’m afraid your data is also consistent with this, this, this. Well, this is really terrible, when you write this to a person, you will either take it as an insult, or you’ll take it, you know, as a wound, or you are horrified, and so on, and there were three kinds of responses to that. From some we never got a response at all, they just didn’t even want to hear about it. There were those who said, would you send me that program, I’m fascinated, I’d like to use it, and the others, the third one, that was the most interesting and amusing one, said, ah yeah, very interesting, but of course this is not possible, the one that the computer sent, because we have another piece of information that excludes it, and they did have it, but they didn’t publish it. And that is the important thing, in other words, their rigorous analytical process was not really systematic and rigorous. They happened to find little driblets, and said, oh yeah, but this we had found somewhere else, and that excludes this one, and so on, and that is very, very useful, and that really... we had quite... made quite a contribution there, in the '60s, and '70s, and even '80s, in this particular area, which was, sort of, the last physical method, you might say, that we employed there.