Why did "Artificial Intelligence" stay intact as a coherent, unified field of study?

Question

Artificial Intelligence is a rather pernicious label to attach to a very mixed bunch of activities, and one could argue that the sooner we forget it the better. It would be disastrous to conclude that AI was a Bad Thing and should not be supported, and it would be disastrous to conclude that it was a Good Thing and should have privileged access to the money tap. The former would tend to penalise well-based efforts to make computers do complicated things which had not been programmed before, and the latter would be a great waste of resources. AI does not refer to anything definite enough to have a coherent policy about in this way.---Dr. R. M. Needham, in a commentary on the Lighthill Report and the Sutherland Reply, 1973

43 years later...

There is already strong demand for engineers and scientists working on artificial intelligence in many of the fields you mention, and many more. But expertise in making real-time systems for controlling trains doesn't make you know anything about robotics. Analyzing human behavior to detect crime has virtually nothing in common with self-driving cars (beyond CS/pattern recognition building blocks). There is never going to be demand for someone with a broad sense of all these areas without any deep expertise, and there is never going to be someone with 300 PhDs who can work in all of them. TL;DR -- AI is not a branch, it's a tree. --Matthew Read, in a comment on Area 51 Stackexchange, 2016

AI is a label that is applied to a "very mixed bunch of activities". The only unifying feature between all those activities is the fact that they deal with machines in some fashion, but since there are so many ways to use a machine, the field's output may seem rather incoherent and incongruent. It does seem to make more sense for the AI field to collapse entirely, and instead be replaced by a multitude of specialized fields that don't really interact with one another. Sir James Lighthill appeared to have supported this sort of approach within his 1973 report on the state of artificial intelligence research.

Yet, today, this Artificial Intelligence SE exist, and we still talk of AI as a unified, coherent field of study. Why did this happen? Why did AI survive, despite its "big tent" nature?

Answer 1

AI is a rather unusual research field in that the label persists more because it represents a highly desired goal, rather than (as with most other fields) the means, substrate or methodology by which that goal is achieved.

we still talk of AI as a unified, coherent field of study

Despite recent efforts in AGI, I don't think that AI is actually a very unified or coherent field. This is not necessarily a bad thing - when attempting to mimic the most complex phenomenon known to us (i.e. human intelligence) then multiple, sometimes seemingly conflicting perspectives may be our best way of making progress.

Answer 2

A lot of the survival power of the A.I. label comes from the popularity of science fiction, which many scientists - computer or otherwise - are big fans of, as are their consumers. Astronomers and physicists, for example, may frown on really bad sci-fi, but I see many of the well-known ones like Hawking daydreaming about things like wormholes and time travel etc. Which is fine - there's nothing wrong with a sense of wonder, as long as it doesn't dupe us into overestimating our success or finding the wrong answers to real-world problems.

Unfortunately, that's a big issue in A.I. research. We watch movies like 2001: A Space Odyssey and Terminator and then set about replicating the fictional technologies seen in them, without even having a hard definition of intelligence. A.I. is a much more melodramatic moniker than say, "Autonomous Algorithmic Pattern Recognition" or some similarly boring label. Because this name is applied carelessly to a wide variety of disciplines, it implies that we have already made significant progress towards replicating advanced aspects of human thought, like consciousness, reasoning, intuition, etc.

In other words, this vague label enables us to fool ourselves into thinking we're a lot closer to perfecting kinds of technologies we see in the movies; the backwards logic boils down to, "because we've chosen to call this odd (sloppy) selection of fields 'artificial intelligence', we must be close to achieving artificial intelligence." The label survives in large part for irrational, human reasons.

I'm not saying that's the only reason, or that some of the other reasons don't have better legitimacy, but this is a big issue that we will have to contend with for a long time to come.

Answer 3

I don't believe that AI as a coherent field has a lesser legitimacy than, say, Engineering. Ignoring for the moment that we're a day or two behind on AI, they're very much alike: Both fields contain a wide variety of sub-fields which stretch across multiple disciplines (although admittedly more pronounced in AI) , in both fields it is mandatory to specialize and in both of them an expert in one sub-field will be more or less useless in a different one (the expert on bridge construction will probably not be very versed in the thermodynamics of AC systems and vice versa). This pattern can be seen in many of today's disciplines - in fact, I don't know if there still is a reputable field, in which a single person can be a universal expert.

You mentioned that the only unifying thing about AI was it's dealing with machines in some fashion - but such a simplifying statement can be made about almost any field. To return to my previous example: the only unifying thing about the various Engineering activities is that they're all somehow involved in the construction of something (be it a flashlight or an aircraft carrier).

AI is a young field and therefore its branches have not yet been established in the sophisticated way that the branches of other fields have, but I would assume that it is only a matter of time until the various differentiations and the corresponding degrees, courses etc. develop.

AI is also growing up in a time where vast knowledge in its related/parental fields already exists and further knowledge is produced at dizzying speeds - and that is as much a blessing as it is a curse. When Engineering was 'created' a few millennia ago (please excuse my ridiculously inaccurate science history lessons) there wasn't much going on in the world of science and so the field grew slowly, with plenty of time to get organized and structured. That is a luxury which AI did/does not have. It emerged in an age of technical wonders, surrounded by scientific breakthroughs on at least a monthly basis and the rise of interdisciplinary science (which by itself complicated things quite a bit). So in addition to organizing itself, the field also has to continuously integrate the large number of advancements made and somehow stand its ground against the outlandish expectations generated by other science's breakthroughs over the past decades and the media (as already explained by SQLServerSteve).

Long story short: it's similar in it's complexity and diversity to other fields and therefore has no reason to collapse - on the contrary, its failure to do so over the past ~50 rather complicated years indicates, that it will further solidify and organize itself in the future.

Answer 4

Because, ultimately, AI is a cohesive "thing". It's an effort to make computers do things that currently only humans can do well. Sure there are many, many approaches and techniques, but there's always been a clear overall goal (although the goal-posts keep getting moved further out, which is a different issue).

As long as there are things humans can do well that computers can't, somebody will be trying to figure out how to close that gap. And those efforts are "Artificial Intelligence".