Daniel G. Bobrow

Artificial intelligence 40 years later

In 1957, the Artificial Intelligence Project was started at MIT. Two young assistant Professors, Marvin Minsky and John McCarthy launched this bold venture after the successful summer conference at Dartmouth in 1956. In that same environment were Warren McCullough, developing ideas about neural nets, Jerry Lettvin, exploring how frogs eyes and brains worked, Norbert Weiner looking at cybernetic models of machine and human behavior, and Noam Chomsky, developing formal models of language competence. Meanwhile, at what is now Carnegie Mellon University, Allen Newell and Herbert Simon were exploring Complex Information Processing-building models of human problem solving. In those first heady years, Artificial Intelligence showed promise in automating mathematical activity, understanding language, using logic to solve problems, modeling human intelligence in an analogical reasoning task, planning, and learning. Back then we thought, if only machines were a little bigger, and we had just one or two more ideas, we could develop theories and prototypes of intelligence that would both inform us about how humans work, and be our partners in solving hard problems. The bad news is that we were over-optimistic. The scale and difficulty of the problems were not apparent to us then (and may still not be now). The good news is that we have made tremendous strides both in our understanding of the theory, and in applying those theories to real problems. IJCAI-97 took place August 23-29, 1997 in Nagoya. A highlight of that conference was an impressive series of invited lectures that provided excellent examples of how far the field of Artificial Intelligence has progressed, stories of some major accomplishments and its major challenges. We decided to invite each of these speakers to submit a paper on the sub.ject of their talk to a special issue of the journal, providing them the space and the time to expand on their subject. The result is this sterling collection of papers. In continuity with those initial efforts, they show how AI has evolved to deal with problems of automating mathematics, understanding and using natural language, using logic to solve problems, planning, and learning. The first set of papers provide a look at how Artificial Intelligence is being used in “applications”, problems embedded in and defined by a real world environment. These

Dynamic Documents and Situated Processes: Building on Local Knowledge in Field Service

Documents play a central role inside corporations as a resource for business processes, and paper has been the primary medium for distributing documents. These paper documents have not always been effective with respect to achieving the goals of the users of the documents, or of the corporations that have designed the processes. As documents are embedded into electronic media, it is possible to rethink their form, method of production and use. This paper is a case study of the redesign of a document-supported process for servicing Xerox equipment. Design and development of the new tools and process involved both the producers and users of the documents, so that they better fit the work practices of both groups.

A wearable and mobile intervention delivery system for individuals with panic disorder

Panic disorder is a serious condition that affects approximately six million adults in the United States per year. Reducing the severity of panic attack symptoms would allow a better quality of life for panic attack sufferers. This paper presents steps towards a mobile and wearable system that aims to help reduce the severity of symptoms experienced by individuals with this condition. The system provides a way to continuously monitor the physiological data of an individual via a wearable device. Users are able to report when panic attacks take place, along with a rating of the severity of symptoms experienced. Reported episodes provide ground truth data to build panic prediction models. The eventual goal of the system is to make predictions about approaching panic attacks and to deliver interventions that help the individual to cope with the approaching episode. We describe a mobile-based intervention that has been developed, which instructs the individual to perform breathing and relaxation exercises. Presently, the system has been utilized in a small pilot study where 10 individuals who suffer from panic disorder reported 29 panic attacks while collecting physiological data, along with the severity of symptoms. We found that out of 15 symptoms the ones with high severity reported were anxiety, worry and shortness of breath. Furthermore, physiological differences were observed between panic and non-panic intervals.

High level language programming environments

Although written by a single author, this report incorporates the views of the entire group that spent long hours in earl y May working towards a consensus view of programming environments. The author wishes to thank all of them for their help. 2 .0 The Mode l Software is often delivered late, is unreliable, is not maintainable and has a host of other well known problems. In order to look a t possible improvements a model programming environment was developed. It was helieved that this model could incorporate most o f the ideas needed to greatly improve programmer productivity. This model is sufficiently general so it is usable even if some of th e needed research results, mentioned later, are not fully developed. Although the use of high level languages is rarely explicitl y mentioned, the underlying assumption is that such a set of languages is needed to interact with the implementation of the model. More about this later. Assumed hardware and software technology circa 1990 was the starting point of our discussions. Although mentioned as a date, i t was used only as some unspecified future milestone-we explicitly did not want to be bound by any near term (within next five years) developments since these would be adequately covered by two of the other three groups at the workshop. We assumed that any suc h system that we proposed would probably be built from scratch, and would not need to have its goals compromised by the reality o f extending an existing system. Experience of several of the members in the group on current day experiments in programmin g environments (PIE [Goldstein], PDS [Cheatham]) gave us first hand knowledge of some of the problems in developing such systems. The basic model consists of a set of objects that are hierarchically related. An object is a collection of entities, where an entity is a se t of (attribute, value) pairs. Attributes are things like author, source code, specifications, date of creation and version number. B y having a two level structure of attributes and entities, the same concepts can be viewed in different ways by different individuals. Thus each user of such a system would have his own perspective of the environment, i. e ., each user would have his own vie w environment as part of the larger programming environment. For example, a given module might have the attributes …