ntm1169@dsac.dla.mil (Mott Given) (06/03/91)
From article <1991May31.221440.2980@cs.sfu.ca>, by edwin@fornax: > How far can expert systems go?? > or, > What, potentially, can they accomplish?? Their accomplishments can be significant. Canon uses an expert system to generate new ideas and designs in lenses design, and is able design lenses 10 times faster. AI tools can be used in conventional programming instead of languages like Cobol to speed up software development; eg. American Express developed an application in 6 months using IBM's Expert System Environment (a mainframe expert system shell) instead of the 3 years it would have required with other languages. NASA is using cooperating expert systems for automated monitoring and diagnostics. The Fusion Group (New York) has created a network management application for a distributed network of UNIX workstations. Ultra-human (i.e. better than human) performance has been achieved by expert systems in some domains by melding the knowledge of several experts and using the expert system for guidance where there are so many factors to be considered in the problem that it is easy even for an expert to overlook some aspects of the optimal solution. A good way for you to get an overview of this is write to AI vendors and get their literature. Some vendors even have free newsletters: ?-consult(user). (from Quintus Corp., 800-542-1283) Aion Update (from AION Corp. 312-380-8870) ARTLines (from Inference Corp. 602-585-3066) AI Interactions (from Texas Instruments) LUCID Moments (from Lucid Inc. 415-329-8400) NeuralWorks Connection (from NeuralWare, 412-787-8222) Software AE Bulletin (from Software A & E, 703-276-7910) > Will they ever replace human experts?? > Will they ever be trusted?? Some references on these topics: AI EXPERT, Nov. 1989, Why expert systems fail BYTE, Jan. 1991, AI's identity crisis IEEE Expert, Feb 1990, AI and expert systems myths, legends, and facts IEEE Expert, June 1990, Validating expert systems IEEE Expert, June 1990, Expert system security "Implementing Japanese AI Techniques: Turning the Tables for a Winning Strategy" by Richard Greene McGraw-Hill Publishing Co. (phone 800-2-MCGRAW), New York, 1990, 266 pages. -- Mott Given @ Defense Logistics Agency Systems Automation Center, DSAC-TMP, Bldg. 27-1, P.O. Box 1605, Columbus, OH 43216-5002 INTERNET: mgiven@dsac.dla.mil UUCP: ...{osu-cis}!dsac!mgiven Phone: 614-238-9431 AUTOVON: 850-9431 FAX: 614-238-9928 I speak for myself
marti@mint.inf.ethz.ch (Robert Marti) (06/05/91)
In article <2948@dsac.dla.mil> ntm1169@dsac.dla.mil (Mott Given) writes: >[...] AI tools can be used in conventional programming instead >of languages like Cobol to speed up software development; eg. American >Express developed an application in 6 months using IBM's Expert System >Environment (a mainframe expert system shell) instead of the 3 years it >would have required with other languages. I have been arguing that using AI tools speeds up software development myself. Still, how do you (or American Express, if you prefer) know that developing the above mentioned application in "another language" would have required the three years you claim? What kind of application was this? And what is "another language", for that matter? Cobol? C? Pascal? An object-oriented language such as C++ or Smalltalk? Lisp? Prolog? A so-called 4GL? HyperCard? Does anyone know of serious scientific experiments which have compared the different languages? (In a SIGMOD Record a couple of years back, someone compared Lisp and C++ but that's all I'm aware of ...) Robert Marti | Phone: +41 1 254 72 60 Institut fur Informationssysteme | FAX: +41 1 262 39 73 ETH-Zentrum | E-Mail: marti@inf.ethz.ch CH-8092 Zurich, Switzerland |