To add AI, or not to add AI?

摘要

The field of artificial intelligence has, in recent years, emerged from the research laboratories of academia and made a large splash in the world of practical software. AI holds a promise for using computer to break new ground, to solve otherwise intractable problems and to improve the quality of current computer systems — it promises a qualitative advance in applications software. However, not everyone would agree with this summary. Alternatively or additionally (depending on your prejudices), adding AI to practical software threatens to wreak havoc on an already none-too-stable technology: all practical computer systems are known to contain errors; many large and very useful systems are quite beyond the comprehension of anyone. They are used as long as they appear to be giving out useful information, but nobody is quite sure how or why they do certain things, and only a small portion of what they generate can be cross-checked for correctness. To add AI, or not to add AI?, that is the question addressed.