Subject:
New Research Study - Practical Aspects of Developing Artificial Intelligence Based Systems
Author:
Richard Ellis
Date:
10/27/2003
To all with an interest in or experience of developing Artificial Intelligence based systems,
Why do AI based projects fail so often? I am often asked this question by people sceptical about the value of AI. My response to them is usually defensive, as I explain (as many of us in the AI community have done) that in fact AI-based systems are out there and working.
But the sceptics do seem to have a point. We go along to conferences and see innumerable papers on "applications" but precious few products that actually make it to market. What is going on? Are AI-based projects particularly prone to failure? And if so, why? Is it something that is inherrent in the nature of AI, or is it something common to the introduction of new technologies? Is it related to the fact that many AI based projects have their roots in the academic world? Is it something to do with the design and definition of systems, failures in project management or decision making, a lack of understanding of the market or the management of suppliers? Or is it something to do with the culture of AI people?
A research sudy has been established to explore these questions. Full details can be found at www.ai-research.org.uk . The research is being run by Stratum, a small technology consultancy, but the results will be made freely available to the AI community.
The study team are actively seeking input to the study. If you have views on the practical aspects of developing AI based systems, then please visit the website and contribute to the study by contacting the organisers or by completing one of the online enquiry forms. All those participating will be kept informed of progress.
Lessons learned from AI-based projects that did not reach their full potential will be of particular value. The personal and commercial sensitivity of such material is recognised by the study team, and you can rest assured that all inputs will be treated in the strictest confidence, and individual projects will not be identified in published material. Anonymous contributions will be accepted.
The study will be supported by a workshop to be held as part of AI-2003, the Annual Conference of the BCS Specialist Group on AI, to be held in Cambridge, UK in December 2003. For further details see the AI-2003 website (www.bcs-sgai.org/ai2003).
Contributions to the debate, either through the study webpages ( www.ai-research.org.uk ) or in this forum, would be welcomed.
Richard Ellis
Stratum
Study Leader
Full contact details at www.ai-research.org.uk
|