[achairapart]

Found this old Wayne Ratliff' interview in a 1984 PC-MAG issue[0].

    PC: What is this “big picture”? 

    RATLIFF: I have to be a little careful about what I answer. It's probably safe to say Artificial Intelligence.

    PC: How would you define Artificial Intelligence?

    RATLIFF: One way to define Artificial Intelligence is "making computers easier to use." However, we don't Just want to make them five percent easier to use, want to make them dramatically easier to use. We are looking for a breakthrough. Eventually, what we want Artificial Intelligence to do is to take over mechanical duties, to free people for non-mechanical things. I want to see computers in my lifetime — preferably in my hand — performing chores in a human, nonrigid, easy-to-use way. I'd like to be able to tell the computer, "Go and total all the checks I wrote in the last 10 years for medical expenses." That's a nonrigid request.

    PC: Do you foresee that dBASE II will be a nucleus for an artificially intelligent system?

...

[0]: https://archive.org/details/PC-Mag-1984-02-07/page/n135/mode...

[achairapart]

I don't know how I fell down this rabbit hole tonight, but here is another interview[0], I think circa 1985:

    Susan Lammers: Have you ever explored the field of artificial intelligence?

    Wayne Ratliff: I was really involved in AI at the start of this business. A little over a year ago, I turned to AI, because I thought that was the future. But I've grown away from it.

    AI has a future, but it's not very immediate. First of all, there's the problem of natural language. If you have a natural-language system, you buy it and bring it home and put it on your computer. Then you have to go through a weeks-long, maybe months-long process to teach it what your particular words mean. The same word has different meanings in different contexts. Even what would appear to be a straightforward word, like "profit," can have a variety of meanings. It needs to be very explicitly defined, based on which business you're in and how your books are set up, and that sort of thing. So this long process necessary for training the machine kills AI, as far as it being a turnkey product.

    But the other side, which is very interesting, is expert systems. My prediction is that within the next two or three years, expert systems will no longer be associated with artificial intelligence. That's been the history of AI: when something starts to become fairly well known, it splits off. Pattern recognition used to be considered AI, but now it is a separate field. That's the immediate destiny of expert systems. I think expert systems are going to be very important in our industry, analogous to vertical applications. 

[0]: http://www.foxprohistory.org/interview_wayne_ratliff.htm

I think it's from this book: https://www.goodreads.com/book/show/2092682.Programmers_at_W...