Multiplied Intelligence Personnel Simulator (M.I.P.S.) Copyright 1987, Michael Reed. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ It has taken over 6 years of research and development to create this system. Many versions have been created in several languages and on a number of different computer systems. Any contributions to help fund the continued research and development of this new form of Artificial Intelligence would be much appreciated. Make checks payable to Michael Reed. By Michael Reed 10308 Metcalf, Suite 342 Overland Park, KS 66212 (913) 341-7412 Press CTRL-BREAK to go to the MAIN MENU ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ In early 1979 I saw my first simple example of what might have been labeled Artificial Intelligence (A.I.). The program was demonstrated on a Radio Shack Model I micro computer. The name of the program was Eliza and for a very short time, I was taken back a little by its ability to respond to whatever I typed. After a bit more time, I was disappointed to find that Eliza was obviously only picking out one to three key words from each sentence and then throwing in a bit of random choice to find its responses. This led to a very boring conversation between the machine and myself. Eliza also never really answered any of my questions or even responded with anything but a simple question of her own. Upon examining the BASIC code for Eliza, it was rather obvious that every response Eliza could give was included within her BASIC code. Eliza could not learn, so soon I knew her too well to enjoy her company or see any real use for this program. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ In 1982 I was developing educational software to run on ATARI computers. In my spare time I was also developing a program called Fred. Fred, like Eliza, was designed to allow me to have a conversation with my computer. In Fred's case though you had to type a question that exactly matched a sentence he already recognized. But, unlike Eliza, if Fred did not have a reply to my sentence, he would ask me to tell him how he should have answered. At that point he added the information as the last line in his program. Then when I asked the exact same question again, Fred knew a response. Because I had to type the sentence exactly as Fred had seen it before, I saw no real use for him. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ In the summer of 1984, I was teaching RPG II in an IBM S/34 environment. It was then that I saw my first application for Fred. Most days I would lecture for an hour or so and then bring my class to the lab. It seemed that, out of an average of 10 students requesting help, only 3 or 4 different questions would arise. The students stood patiently in line waiting for me to answer their questions and I found myself repeating the same short simple answer to a series of students. These question/answer sessions reminded me of my Fred program. I developed a simple version of Fred in IBM BASIC and allowed students to sign-up to use the new system. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ Students who registered could call up Fred and ask him their question before standing in line to ask me. If Fred knew the answer then they went back to work with answer in mind. If Fred could not answer, then the student stood in line as always. After I told a student the answer, I would also tell it to Fred so he could answer that question from now on. Again, the problem that Fred could only answer if you typed a question exactly as he had seen it before, showed its face as a difficult-to-work-with feature. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ The year 1985 brought me back to the micro computer world. I purchased an IBM PC/XT and began work on Dave. Dave, unlike Fred, could recognize questions with similar meaning to any question he had been asked before even if they are worded quite differently. After proving that my new technique worked, I began working on Kipp. Kipp, like most other versions before, was written in BASIC. Although a large BASIC program Kipp had some wonderful features. The real feature that separated Kipp's ability from Dave's was that Kipp could be used on-line. I had a second phone line installed and modem users from all over my area began calling Kipp 24 hours a day. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ As users chatted with Kipp, he recorded anything that he did not recognize. This allowed me to continually add the new information to the system. After one year of on-line conversations, Kipp had registered more than 800 users and could discuss seven different topics: help, phone numbers, electronics, computers, wines, first aid and conversation. In 1986 I decided that the technology had been proven, and it was time to shrink the size and increase the speed of the program. 8088 assembler seemed to be my best answer. Everyone who saw the new version, which I call Multiplied Intelligence Personnel Simulator (M.I.P.S.), was amazed by its incredible speed. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ Kipp averaged 20-second response time while the new system never seems to pause. M.I.P.S. also displays its information in three separate windows. The top window is used to type in your question or statement. The bottom window displays the sentence you typed and how M.I.P.S. read the sentence. The large center window displays the system's response to your sentence. This windowing set-up allows me to type in my question and then examine the bottom window to see if the system understood me. If M.I.P.S. did understand then I assume the response I received was correct. If M.I.P.S. did not understand, then I would type WRONG, and my question is recorded so that it can be taught to M.I.P.S. at a later time. It is easy for me to see many uses for my system now. It just needs knowledge about subjects that would be beneficial to its users. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ 1988 will be a year of learning for M.I.P.S. If you have knowledge about any topic and a real interest in A.I., then become one of the original M.I.P.S. teachers. Once you build a knowledge base within M.I.P.S., that information can be given or sold to any other M.I.P.S. user. Teaching M.I.P.S. requires no previous programming experience. If you can ask a question and then give the answer, you have all you need to teach this amazing system. You don't have to know about nouns, verbs, synonyms or any parts of speech. You can even teach M.I.P.S. to work with most any human language that can be typed using the computers keyboard. ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ» º HISTORY BEHIND M.I.P.S.º ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ If you need a copy of M.I.P.S., have questions or comments, please call or write to me at the address below. Feel free to make copies of this system for your friends. There is no charge for this program. Michael Reed 10308 Metcalf, Suite 342 Overland Park, KS 66212 (913) 341-7412 Any contributions to help fund the continued research and development of this new form of A.I. would be much appreciated. Make checks payable to Michael Reed.