Personalized learning based on algorithms and inferential statistics - My fear of drowning the voice of the learner.

(image credit to: Andrew K Miller)

Personalized learning.

This is coming up with might. Zuckerberg-Chan are doing it, Microsoft is doing it, Google is doing it, hell - Bertelsmann is doing it ... all the "big guys" are getting with might behind the paradigm of personalised learning. Many seem to feel compelled to employ "big data" in order to foster the personalisation process. Zuckerberg-Chan's approach probably being the most thoughtful of them, still may lack some fundamental awareness - but let me come to this later.

(I will leave the aspects of data security and danger of misuse aside. You can read about it in the case of Google here.)

Yes, we did understand that standardisation in education is not a good thing. Learning is about people, about challenging the creative mind. Learning is (or rather should be) about curiosity, discovery, experimentation, imagination, engagement and passion.

Here is an official definition of personalized learning:

“Personalized learning is tailoring learning for each student’s strengths, needs and interests — including enabling student voice and choice in what, how, when and where they learn — to provide flexibility and supports to ensure mastery of the highest standards possible.”

- iNACOL

The most important part of this definition is, in my eyes :

"..including enabling student voice.."

But then I listen to panel talks of e.g Dr. Jörg Dräger (Bertelsmann) who promotes the "big data" approach, which is to collect historical data, achievements and all possibly reachable data about a person in order to apply "sophisticated algorithms" that can infer where the student has his strengths and what his predicted success rate will be on a compiled and suggested study path.

Is there actually room for the student to say, no ? I mean a predicted 62% success rate when I take a certain educational path which includes the "right" topics, compiled from my past achievements and involvements. Isn't that an easy one ? 

Maybe we should step back a second and try to understand that when a person is good at something, it does not automatically mean this person is particularly passionate and interested about it. Just talented.

A story told by Sir Ken Robinson comes to my mind, where an accomplished Pianist who filled the largest theatres in the world gets asked after another successful concert: "did you enjoy it" ... and the answer of her was: "not particularly, no" ... "why are you doing it then?"

These two questions changed her life. It turned out she did something nobody doubted she would do, because she was good at it. Not even herself doubted or put too many thought into it. Her parents were both musicians and she was a natural, but her passion, her passion was books. 

On that day she closed the lid of the piano and never opened it since. She took a job as an editor, is underpaid but says, has never been happier in her life. She is doing something which resonates with her soul, she lives her dream.

Using "big data" means looking at the history of someone, what the person did, wrote about, achieved, scored in school etc .. Then using this sources, to break down the personality of someone based on demonstrated abilities, in order to categorise this person and tell him/her what the path of learning should be, because of the greatest likelihood of success.

Is anyone else thinking about the movie "Divergent" ? The danger to categorise students, the danger of sorting them into factions comes to my mind. 

What would inferential statistics and "sophisticated algorithms" have told us about Albert Einstein ? He was known as a slow learner in school and an under achiever in Math ? "Albert, let go of Math and Physics, you will not likely succeed there?". I think we can be happy that the idea of using "big data" was not around to assess his personal interests. Nobody could have known them. Thankfully - no convincing "big data" assessment was there to discourage him from making his thought experiments.

Why do we think everything needs to be shown to the learner instead of helping on the individual quest to find his/her own way and figure it out ? I think there is the malady. The big smart guys seem to know better what you want and can do, here is the data, the data does not lie. If you don't follow our advice you will likely fail.

To make it an even stronger message: Dr. Dräger does not hesitate to announce success. The predictions that were made so far show that the system works. Why should that impress? If you tell students they should take these topics because they have the highest success probability there, based on what they showed they could do best in the past, this seems to be not much of a surprise.

A surprise for me would be if that always matches with what the students actually want, feel and are really passionate about. What makes them happy.

It is dangerous to think you know because you have enough data, you don't. Your data does not look into the heart and soul of the person. Start from the person, not the data about a person. Machines do not measure creativity, feelings, imagination and dreams.

Help the students to find their individual paths, interests and passions. Provide the creative space and time to experiment and imagine, then show the data if you must. It can help, there is no doubt about it, but that is it. It should be intended to help, to be one color that can be used to paint the whole picture. But most importantly, help the students to paint their own picture, do not try to paint it for them!

So this is my fear, that we drown the inner voice of the student, because we do not listen well, we rather tell. We do not help to find a path to happiness and fulfillment, we show the path with the highest probability of (what we think is) success.

Machines are good at compiling data, they are good at crunching numbers, at inferential statistics - they can make predictions based on historical data. But when it comes to the human mind, I think the complexity is not going to be calculable by algorithms any time soon (if ever). Unless machines understand the concept of feelings, humour and grief, shame, love and joy I think that the predictions of a machine simply lacks what it takes to profile the whole human being. 

Besides I would not like any machine or sophisticated algorithm to calculate what is good for my children, what career path is best for them. I know that they are flighty, change their minds and preferences often and are eager to discover the world and their passion on their own. A machine may be able to predict what they are likely to be good at, but it can't predict what will make them happy.

More than that, an assessment cannot be a one off, it would have to be a continuous process. This is what life is, what development in a person is and should be: a continuous discovery of new possibilities and capabilities, questioning old wisdoms and challenging the status quo.

Being familiar with the software industry for 25+ years, I had the chance to witness amazing progress. It is possible to create Software that helps students to explore and to find the own path, I know it - we did it. We need not show a path, but be a modest aid for learners in finding or creating their own.

For my next posts I would like to invite software makers to re-think with me together what a learning software should be. A learning software which helps to experiment, helps to learn how to learn, without an emphasises on the what.

We are not talking about memorising here. What I am talking about is the FIRST "4th Generation Learning Software", which aims to be an aid for learners to explore their creative potential and interests and to discover their passions. We created this software, and we want you to be able to follow our example creating your own version of it, improve on it when you can. Why ? Because our children deserve the chance to be happy, this is how I would define success. What about you ?