Keith Douglas
Last time I continued our analysis of artificial neural network approaches to AI, doing a little bit of history and explaining the “inexplicable” label that is often applied to them. This time we are going to continue on the same general topic, but address one way in which the problems discussed then and the time before could be addressed. This is professionalization.
Professionalization
Certain jobs, for lack of a better phrase, involve different degrees of training, education, supervised experience, etc. Relatedly, to use certain titles, one must have certain sorts of responsibilities, etc., often requiring similar considerations. These often legal restrictions are variable across jurisdictions. Here in Canada, similar broad rules apply in most provinces for most domains but the details vary.
One area where they vary quite a bit is in clinical psychology domains and more generally in psychotherapy; I am told that Quebec has less educational requirements. Some fields, including psychology and chemistry, have both professionally organized fields under suitable legislation and others which are not so covered. Further fields so covered are many healthcare professionals: physicians, nurses, pharmacists. Engineers are as well; even using the title “engineer” is at least morally dubious without being a professional.
We will return to this important matter a bit later on. Teachers (but not college instructors or university faculty), lawyers and accountants are also usually included. Finally, it is important to realize that professionalization does not seem to guarantee what one might call “intellectual respectability” or more generally the avoidance of pseudoscience and similar: Chiropractors are often professionalized as well. Nevertheless, professionalization (for all fields) in this sense nominally involves appropriate course work inside and outside a classroom, often some sort of professional ethics and practice background, and examinations or other peer evaluation.
With this cursory background in mind, let us rhetorically ask: “What computing fields are covered by this legislation and professional policy?” The answer is usually one and rarely two. If one is a computer engineer, one can certify as a professional engineer. Computer engineers are generally responsible for hardware design and implementation (e.g., microprocessors). I have lost track of my friends who were academically trained as such; many of them were not certified as PEng because they did not wish to do so and it was not required of them. Some of them worked at the hardware/software interface (firmware, device drivers) as well, and there is even less in software.
Software engineering proper is very new. It did not exist when I was a student 20 to 25 years ago. When the term was used, particularly Carnegie Mellon’s famous Software Engineering Institute, these programs of study and courses generally did not lead to professional certification. Regardless of the state of the existence of various forms of professional engineering related to computing, I know of no context in which there is any legal obligation to have one. Additionally, and important in what follows, none of this directly pertains to security or privacy. There are, however, industry-created certifications in these domains but these are not coordinated through legislation and policy, at least here in Canada.
I would like therefore to move on to present a case in favour of professionalization of computing work.
For Professionalization
The case for computing professionals comes from two distinct angles. One is the danger to the public or to an employer. The other pertains to the danger to the employee or worker. Let us take each in turn. The harms are real. Software, for example, if exploitable or even used inadvertently in dangerous ways, leads to data losses, monetary impacts, health problems (e.g., via ransomware which hits hospitals).
The expertise is known, too: There is a large body of knowledge which exists on how to improve the security, general reliability, and privacy of software. Relatedly, there are recognizable curricula at universities. As yet, however, these do not generally include ethics or professional practice. The proposal to professionalize would likely require their inclusion. The public and the employer can thus rely on a body of knowledge to some degree.
As for the employee or worker herself, professionalization helps her too. She would have the backing of peers when having to confront the employer with refusals to follow regulation or policy, for example. She would have the ability to say “no” to professionally inappropriate activities and have the legal and moral backing. Currently, computing professionals often feel they have to release at any cost or danger, and that their jobs are threatened by security and accessibility aspects to software. Professionalization would not only give them the right to refuse this, but, under certain circumstances, the obligation to do so. For those who want to re-skill or add to their toolbox (e.g., an economist wanting to write analytical software) there would be a clearer path and less possibility for debate over whether someone is qualified.
Let us look briefly at a case against.
Against Professionalization
I will present three prongs to the case against. First, it is quite difficult to know what subfields would be subject to the professionalization. Deciding who is legally and morally responsible for certain aspects of computer systems is going to require a very rigorous analysis of how they are put together. There is effectively no user-level system that is manufactured in any sense by one company. Moreover, in the software space at least, understanding boundaries, parts, and wholes (what philosophers would call the mereology) is hard and contentious. Even something seemingly simple like version numbers has not reached consensus.
I regard the second major challenge to professionalization to be one of “grandfathering.” When engineers were professionalized in Canada, engineering degrees from a recognized university program were a necessary component with three exceptions. These were railway engineers, and two kinds of steam engineers: stationary and mobile. By modern lights these were closer to what we would call technicians (some of which are also professionalized). That aside, there are thousands of Canadians working as software developers, information (or cyber) security professionals, various forms of so-called architect, quality assurance specialists, testers, UX experts, technical support personnel, networking technicians, and beyond. Which of these are covered? How do they relate to the new regime?
The third challenge is related to the second, especially as I said “Canadians.” Canada is an attractive place for immigrants, many of whom bring trades and professional qualifications with them. However, many have trouble re-upping or retraining to the Canadian standards, and so by introducing another area where qualifications are official, we would perhaps lose many worthwhile immigrants.
What do you think?