Interested in the sciences and the humanities? The emerging field of artificial intelligence (AI) spans the intersection of psychology, informatics, and philosophy. And because AI students are trained in such a rich multidisciplinary environment, they have excellent career opportunities.
To make a thinking machine is one of humanity's oldest dreams. And since Allan Turing?s 1947 lectures on AI, programmable computers seemed to be the best way to go. Expectations were extraordinarily high in the 1950s and '60s, but without major breakthroughs, the whole subject lapsed temporarily into obscurity.
Now, advances in the cognitive sciences that are improving our understanding of the nature of intelligence, memory, and perception from the biological perspective, coupled with the ready availability of ever-faster computers, are creating a second spring for AI--and a mecca for multidisciplinary scientists.
Both AI and cognitive science rely on scientists trained in many disciplines, and--as scientific fields go--they overlap considerably. Accordingly, they often come in academic twin packs, as is the case in Utrecht, the Netherlands. Here, Utrecht University?s Cognitive Artificial Intelligence (CKI) undergraduate and graduate courses are based on sound training in mathematics and logic, but at the same time incorporate neurosciences, linguistics, psychology, informatics, and philosophy.
A brief glance at this curriculum suggests that it is exceptionally demanding, a conclusion seemingly supported by the generally high drop-out figures experienced by all the AI courses in the Netherlands--on average only half of second-year students pass their final exams. These figures were published last month in a first-of-its-kind report on the quality of AI courses in Dutch universities by an independent commission chaired by Jan van Eijck of the National Research Institute for Mathematics and Computer Science (CWI).
But the challenges of the interdisciplinary curriculum are only one reason for the high drop-out rate. According to Noor Blaauw, a study advisor at Utrecht, the CKI students are also being snapped up by hungry IT companies--even before they have their diplomas. This view is supported by the commission?s report, which states that the ?most important reason [for the high drop-out rate] is the high number of side-jobs."
But at the same time, the commission praises the generally high quality of AI courses in the Netherlands (see Box 1, below). Although each of these AI courses offers a first-rate curriculum, each also establishes a different disciplinary focus and adopts different teaching methods, says the commission. For example, although Amsterdam?s AI courses are more closely aligned to the computer sciences and Nijmegen University offers an AI course with a strong psychology element, it is the philosophical component that makes the Utrecht course unique. ?Philosophy as part of the AI course attracts people with broader interests" observed Blaauw. And it also attracts women. In stark contrast to the situation in most of the information sciences--and to other AI courses around the country--about 20% of the students in the CKI programme are women.
While broadly focused AI curricula can be a real challenge for beginning students, they turn out to be a major advantage when the time comes for those students to enter the job market. According to Blaauw, AI-ers have not been greatly affected by the recent economic slowdown, particularly in comparison with their peers taking more traditional IT programmes. Nevertheless, a sound understanding of the IT components of AI, such as knowledge of programming languages such as C++ and Java, remains a must for many of the positions outside academia. ?It is hard to think of a field where our graduates are not welcome," Blaauw asserts.
AI students from universities that have a more applied focus (Maastricht, for example, which focuses on mathematics, or Gronigen, which favors physics) make the jump into industry even more frequently. Students from Nijmegen or Utrecht, however, tend to stay in the academic world, at least for a while, say docents from these universities. One reason for this is that AI research is still relatively young, which means that the questions being addressed are fundamental and also that the field is branching and rapidly developing. Moreover, AI students enter a highly mobile and international community of scientists. Says Blaauw, citing examples in Washington, D.C., Melbourne, Berlin, and Rome, ?an increasing number of [CKI] students spend their final year abroad."
However, AI has been losing its purely academic innocence. "We used to joke that AI means 'almost implemented,'" Rodney Brooks, the director of MIT's Artificial Intelligence Laboratory said in an interview in Wired magazine in March 2002. That is changing. As AI enters its new spring it is leaving chess--?the Drosophila of AI," as Alexander Kronrod, a Russian AI researcher called it--far behind. With many new potential fields for AI applications opening up (see Box 2), basic scientists are being lured farther and farther away from their ivory towers.
Box 2 - Current Areas of AI Research:
Adaptive Learning: management; resource allocation; logistics
Text Parsing: Internet search engines; job search databases
Pattern Recognition: forensics--e.g., detecting credit card usage
patterns that could indicate fraud
Expert Systems: medical diagnostics
Speech Recognition: automated information services
Natural Languages: still in its infancy
These changes are likely to influence the universities? curricula and, depending on their area of expertise, shift them towards more applied subjects, says Paul Kamsteeg, researcher and docent on Nijmegens? AI course.
In addition to the vibrant job market for graduates, a second change may provoke interest in AI and the cognitive sciences among many who had not seriously thought about them before. With the beginning of the new Dutch academic year this fall, all universities in the Netherlands will introduce a Bachelor-Master's-structure aimed at making the country?s higher education system more transparent and globally compatible. And although the cognitive sciences are enjoying a stable influx of students, IT courses are not doing so well. The new degree structures will allow students to switch disciplines--or even universities--more easily, opening the field for possible vertical entries from mathematics, psychology, or informatics at the postgraduate level.