When we talk about the future of AI in education, a lot of time it seems like we’re talking about one of two things: 1) adaptive assignments that try to get us to the point where we ‘know’ something as quickly as possible, and 2) technologies that will take care of the gruntwork for teachers so that they can spend more time focusing on the students.
These are both laudable goals. That said, besides positing the student as a passive agent in the learning process, neither presents a radical new understanding of what computers can do for the classroom. You don’t necessarily need AI or machine learning or other ‘smart’ technologies to save time or organize learning in an efficient way. We’ve heard all these ideas before even when computers weren’t particularly intelligent.
If anything, talking about AI in this way impoverishes the digital learning paradigm, which also has put its focus on students’ ability to create and do and communicate in new and exciting ways. Educational technologies provide the most meaningful service when they help us discover something new and learn from it.
But there is one area that could deeply benefit from AI in precisely this way – by providing learners with the tools to discover something new and learn about it. In particular, it will help learners learn about themselves, the most important subject of all. This area is in the developing practice of learning portfolios.
McMaster University provides a good definition of learning portfolios on its website:
A learning portfolio is a purposeful collection of student work that exhibits a student’s effort, progress, achievements and competencies gained during a course or time in university.
If you’re a student, learning portfolios give a you a way to reflect on your progress and showcase their work. They can likewise serve as evidence for parties interested in your skills and abilities – especially for new or emerging 21st-century skill sets for which there may be no suitable assessment framework. They offer a more direct look into what you can do than what someone might find on a resume or in a school transcript.
And while McMaster University is promoting them among its university students, learning portfolios have found success at all levels of education, including junior and senior high schools – and even elementary schools.
Learning portfolios can be compiled for virtually any subject. For some good examples, Portfolium is a website that hosts student portfolios in subjects ranging from culinary arts to physics. Exploring that site can be a great way to understand how student skill sets are changing, and how learning portfolios can serve to demonstrate them in the way traditional transcripts and resumes often fail to do.
Learning portfolios can include ‘classic’ documents like essays, presentations, or project descriptions. Or, as in the portfolios used by graphic designers, artists, musicians, and the like, they can include direct demonstrations of skill – designs, paintings (or digital copies of them), or audio recordings. Programmers, likewise, use public profiles on Github and similar sites to showcase their programs and scripts. Essentially anything can be included if it demonstrates some kind of learned skill.
But learning portfolios can involve other forms of documentary evidence as well, particularly video or images. These are often included as a way to document projects that involve objects or machinery or other things that aren’t so easy to share online. They can also, however, help document an experience, such as a trip or a journey, that is part of a larger learning pathway.
And this latter use is only becoming more relevant in the age of the smartphone – and the attendant ubiquity of digital photgraphy.
A few years ago, people started talking about how the “camera is the keyboard,” that for people using smartphones and other mobile devices, the primary means of input was moving away from typing to taking photos. Anyone who uses a smartphone can understand why. Taking pictures on mobile phones is constantly becoming a better experience. Not only are the physical cameras getting better, but the software that supports them is also improving. Meanwhile, typing on the onscreen keyboard, while improved slightly by better predictive text and better auto-correct, hasn’t gotten much better in the years since the iPhone launched. Even now, it’s a weak part of the experience, comparing unfavorably to the Blackberries it replaced over a decade ago.
This, combined with a generation of smartphone users who have become used to interacting through visual conversations on platforms like Snapchat and Instagram, have meant phone cameras and images have become more fun and engaging. Meanwhile writing text on the phone compares less and less favorably to these things, and is increasingly dominated by by high-context, image-supported memes.
The increasing convenience and popularity of phone photography has already had an impact on education – as it has crept its way into many areas of our lives. When we’d make copies in the past using a Xerox machine, now, in many cases it’s completely serviceable to simply take a picture of a document with a phone. And, of course, it can add useful detail to a learning portfolio in the ways described above.
But digital photography has an even greater potential to shape learning portfolios, thanks to recent developments in machine learning.
One interesting word that the McMaster University website uses to describe its learning portfolios is “purposeful.”
While many people don’t think about it in this way, purpose is really central to the whole assessment process. Assessment and evaluation aren’t just about grades. Grades and point systems exist, rather, so that we can think about what we need to study if we want to improve. When it comes to well-established subjects like Latin or English grammar, grades have been very helpful in allowing us to identify weaknesses, and targeting areas to work on.
Learning portfolios help us think about purpose in a broader sense – or when subjects are new or curricula are not particularly well-established. When working with a learning portfolio, the student needs to answer the following three questions: 1) What do I need to learn, 2) How do I learn it, and 3) How do I show that I’ve learned it outside of a traditional testing structure?
But of these three questions, the first can be extremely hard. Of course, if you’re goal is to become a computer programmer, maybe you only need to figure out which programming language is most popular or useful to a particular purpose. After this, you can proceed on to questions 2 and 3. But if the learning process involves the student engaging some more existential questions, or if the learning process involves softer skills like leadership or community building, then it requires no small amount of self-reflection and self-understanding to even begin answering this first question.
In many cases, finding purpose is an ongoing process and can’t simply be completed at the beginning of a project. Usually the process benefits from a mentor who can help the student think about these his or her purposes.
But it can also be enabled with new machine learning tools – in particular, those related to image recognition – which can be applied to digital images already being put into learning portfolios.
Or it can be used at the beginning of the process by examining the images we’ve already taken. In this way, machine learning offers us the chance to discover learning objectives based on the way in which we’re already living their lives.
For example, a portfolio project could begin with the selection of images from a student’s personal photo album. A trained image recognition algorithm could then go through these images and label them with what’s inside them. The service could then link these photos to other data sources, such as geographical data, historical records, or even vectorized word corpora, and suggest points of contact between these and the images.
These connections could be shaped by subject or skill area that the student wants to study. For example, if the portfolio is for some kind of history class, it could seek to tie these images to date information, other historical records or pictures from an archive service.
It could also look for interdisciplinary connections, analyzing, say, the student’s engagement with plants, animals, pets, or horses, and help the student find links between this area and historical research. It could lead the student to study the way humans have engaged with the local environment over time.
In STEM subjects, image recognition could help the student find practical applications for the various problems, theorems, and so forth that he or she is studying in more abstract terms. For example, it could help the student apply basic physics principles to pictures he or she took at a baseball game. Or it could help the student develop his or her own problem sets to practice with – using finding geometry problems in local buildings or where the student spends his or her time.
And in language learning, it can help the student identify vocabulary words or other language-learning objectives simply based on items it finds in pictures. The image recognition algorithm can analyze, not just the objects that are incidental to your daily life, but situations, and possibly relationships between you and others, as well.
I hope to explore these examples, and other applications, in future articles. But it’s important to note here that this is no mere research tool. By helping students find connections between disparate areas of life – using the now ubiquitous medium of photography as the bridge – they can in fact be developing their own paths through course and subject work, thinking about their own purposes, and targeting skills that will be most useful to them in their lives.
Rather than set his or her purposes at the beginning, digital images can help students set smaller learning goals. These will broaden and deepen the more that that student interacts with the portfolio system. And then, these, in turn, would tie into broader learning objectives as the process unfolds. In this way, larger learning objectives will emerge from the simple documentation of the student’s life.
Besides helping students discover a unique perspective on various subjects, it would teach the student about the valuable role self-reflection, self-listening, and self-knowledge play in successful goal-setting. This in turn would students better at setting goals overall.
Of course, all of this is already possible possible with extensive self-reflection through keeping a journal or through discussions with an engaged mentor. But the emergent portfolio recognizes that taking photos with your phone has become a far more common activity than these other practices – and that image recognition and related data processing open up new and exciting perspectives that were hidden from us before we had them.