Ask most casual observers where artificial intelligence technology and expert systems have made a real impact, and they would name Japan and nowhere much else. But much closer to home, across the channel in France, expert systems are quietly making rapid inroads into business and commerce.
In contrast to the rest of Europe, which remains largely sceptical, indifferent or downright suspicious, les francais adorent l’intelligence artificielle. An eight-day expert systems conference, held last month at Avignon’s Parc des Expositions, proved conclusively that, in terms of research, development and implementation, France has a coherent grasp of an area which many still dismiss as a distant, 21st century concept. Broadly speaking, the artificial intelligence-based research currently taking place throughout the world, can be divided into two specific areas: the development of natural language communications, and the construction of increasingly sophisticated expert systems. The former is the intelligentsia term used to describe attempts to produce a man-machine interface which enables the user to enter instructions in his mother tongue.
Context Progress in this area has been notoriously slow, hampered less by technological considerations than by linguistic difficulties, with specific stumbling blocks proving to be the enormous range of meanings that can be attached to a single word, depending upon the context in which it is used. Although – as in most areas of AI research – Japan and the US lead the way, notable French steps have been made: Cap Gemini Sogeti, for example, now has an information processing module, Texnat, which automatically indexes key words and phrases. France is also contributing to the European Community-funded Esprit programmes, where increasing amounts of time – and money – are being spent on developing language-based products, and is involved in another EC-based project, Eurotra, which has been set up to develop a multi lingual translation system. Meanwhile, significant research is being carried out in French universities, in particular at the laboratory for the automisation of language and information, attached to the Centre National de Recherche Scientifique. It is in the implementation and development of expert systems, however, that France really out-strips her European counterparts.
Despite the findings of a recent survey which revealed that some 61% of potential French users are still undecided about the role expert systems could play within their firms, conference chiefs put the amount spent last year by large French utilities and companies on expert system development at around $50m – about 7% of the figure spent over the same period and for similar purposes in the US. For existing users – some 11% deemed expert systems to be vital, along with 25% who described them as useful, but limited, – the majority of the applications seem to fall into the fault-diagnosis category, with the likes of Elf Aquitaine, Renault, Bull, Thomson and Merlin Guerin pioneering development. Existing examples include Diva, the trouble-shooting system for large machinery developed jointly by Alsthom, Electricite de France and CGE, Cap Gemini Sogeti’s Sitere, implemented by Renault, and Framentec’s Maintex, a fault-detection tool installed in some 25 industrial sites across the country. Second in the user league comes the transport sector, where expert systems are now used to perform a number of additional functions, ranging from vehicle maintenance to traffic routing. Prominent applications include the Gesp system used by the SNCF railway network to plan route modifications in the face of floods and derailments, and the Rufus system developed by the Regie Autonome des Transports Parisiens in conjunction with Cognitech, which is used for the detection of break-downs. Within the aeronautical and space industries, development has been inspired by the Hermes space ship project; Airbus Industrie employs an expert system within the A320 cockpit, Matra has pioneered a fault-detection product for satellite stations, wh
ile TITN has developed a system – long overdue its seems – to lighten the load of the air-traffic controller. Another big French user is industry – be it of the computer-aided engineering style or the more traditional production-line brand. Particularly celebrated is the Marcel Dassault Elfini system, used in aircraft design, with other examples including Cap Gemini Sogeti’s system for optimising the manufact-uring process of gruyere cheese, and a similar product, but designed for a variety of different manufacturing applications, developed by Thomson-CERT.
Ovarian cancers Within the medical sector, expert systems are used as diagnostic tools – Cognitech has a product designed to identify ovarian and other cancers – and as interpretat-ive aids, along-side X-rays and image scanners. Businesses and financial institutions also employ expert systems for a range of problem-solving, client assessment and internal personnel purposes – notably the Ifodiact-developed Prometheus system, Ordimega’s Preface-Tableur and Preface-Expert, and the GSI-TECSI system developed for the insurance arms of building societies to provide speedy client bills of health. Also-rans are security and agriculture, where expert systems are now being used for myriad puposes ranging from avalanche predictions to long-range weather forecasts. One cautionary note was sounded by the conference organisers, who warned against the dangers of hyping artificial intelligence technologies, and alienating users in the process. Expert systems needed to be integrated gradually and carefully with existing tools, they argued, if the artificial intelligence mystique, which still hinders wide-spread adoption, was to be convincingly dispersed. For delegates, however, it was surely a case of preaching to the converted.
