Thursday, November 10, 2016

“This House Believes Artificial Intelligence Could, Should and Will Replace Teachers” - The future of mass higher education.

This post was written by invitation for the Faculty of Science and Health blog in Dublin City University - 

“This House Believes Artificial Intelligence (AI) Could, Should and Will Replace Teachers” is the topic of a debate at the upcoming at the OEB conference in Berlin later this year ( Often when I ask lecturers the question, “If computers could replace lecturers, should they?” I get a negative response justified by various quite valid arguments that teachers will always be required for one reason or other. However, the question they are answering is not the one posed. It ignores the “if” at the start. As an engineer I have always assumed that it was my job to improve the world by making our work more efficient and reducing waste. Sometimes this might include the disappearance of certain professions, but looking back in time it seems that this disruption, although unpleasant for the disappearing profession at the time, was best for society in the long run. So if you are in agreement with that general principle the answer to the question has to be “yes”, computers should replace lecturers if they can.

But before we give up the ghost and let management replace us with computers, there is the question of whether they are capable of replacing us. In the same spirit of misreading the question above, lecturers are loath to admit that computers can, to any great extent, do the tasks that we do, well enough to replace us. This is the natural inclination of any profession to protect itself but it may not be the best strategy if it ignores the changes that are undermining the profession and inhibits the its ability to adapt.

So, to what extent is information technology capable of replacing lecturers? You might say that the first big scare we got was from the MOOCs, Massive Open Online Courses. The demise of higher education as we know it was being heralded by the sight of tens of thousands of learners taking courses from rockstar professors. It was cold comfort to the profession that the MOOC drop-out rate was high, or that the teaching was simple, or interaction between learners limited (and non-existent with the professors), because large numbers were still learning some very useful stuff.

However, we had an ace up our sleeve; assessment and accreditation. Despite our claims of lofty learning objectives, we know that young people come to university to have a good time and to get a certificate that will get them a decent paying job. Employers, parents and students themselves trust us to maintain good standards and not be handing out certificates to anyone who is willing to pay the fees.

So we’re OK then? Maybe not. A number of years ago the presidents of US universities were polled on what they believed would cause the most change in higher education in the future. Sensibly, they suggested it wasn’t MOOCs. However, they did identify the idea of unbundling, or the separation of learning from assessment, which allows learners to learn as they please and to submit themselves for assessment whenever they are ready. If implemented in universities it would unleash a wave of innovation in learning both inside and outside universities as learner seek the most cost effective ways of learning wherever they can find them. Since then this has started to happen. Universities in the US are offering challenge examinations with credits attached. People predicted that the prestigious institutions would not get involved in this but MIT is launching micro-masters degrees where you can study the module for free in the form of MOOCs and then pay for assessment to get the credits.

It could be even worse. These free and cheap online courses may get a lot better. Progress is being made in Artificial Intelligence (AI) and it may have a significant impact in several areas. Deep learning, an AI technique, can monitor the activity of very large numbers of learners in order to optimise and personalise learning pathways (including remedial activities) for individual students. AI has also made progress in creating “bots” who can act as first line advisers to learners who are in difficulty. (Some recent tests found that the only way students could identify bots was by their response speed).

OK, so we may get replaced, eventually, but we’re smart and we can find other useful thing to do (like research). And the universities will still act as assessors and accreditors so they’re OK. Well maybe. Employers have mixed feeling about the education young people receive at university. On one hand they complain about the lack of key skills in graduates but on the other hand they find universities to be largely trustworthy in maintaining standards and they particularly like their ability to help them evaluate job applicants. University degrees are effectively a cheap selection tool for employers. However, they are not so cheap for the candidates or the state who may subsidise the process.

But what if employers found other more efficient ways of assessing what a job candidate knows or can do? Recently a major international management consultancy firm announce that it would no longer require a university degree from new employment applicants. Whatever, this implies about the value of degrees, it certainly indicates that they believe that learning can come from other sources and that they have the competence to evaluate such learning. In domains where there are dire skills shortages, individuals are turning to alternatives such as “boot-camps” and free online courses with added fees for assessment. Many of these are creating “alternative credentials” with electronic certification which can be displayed online and, more importantly, examined in detail by potential employers who can drill down to see much more detail on the contents of the courses and performance of the candidates. Should such credentials gain the trust of employers the associated course may well prove to be more attractive to school leavers who need to balance the pleasure of the “college experience” against their employability and the total cost of their education.

So to conclude, there is every reason to believe that technology may be able to replace a lot of what we as teachers do in the near future, and the trust the public rightly places in our institutions may not be enough to protect the profession. We may be heading for a time where the need for teachers in much less and their role is much different. Not only do we owe it to ourselves to prepare for such a possibility, we also need to admit that if it provides better, cheaper and more accessible learning for the public, it is to be welcomed.