My Journey into AI Part 2



In Part 1 of this series, I discussed my foray into Artificial Intelligence, bots and the impact these could have on education. After interviewing Oli Trussell on #Episode 31 of the podcast, I was intrigued by Artificial Intelligence, bots and the impact these could have on education. I started the free online course #ElementsofAI from the University of Helsinki and completed Chapter 1. As we look forward to interviewing Aftab Hussain, the ILT Manager from Bolton College who is part of a team that has won a heap of awards for their use of chatbots in education. With all this correlation, I signed up for the course Scott recommended immediately!

As a related side note, I read an excellent article by Mark Steed, Artificial Intelligence, Ethics and Education, in Digital Strategy magazine. Mark is Director of JESS in Dubai and makes some amazing points around the ethics of AI and its place in education. He notes,

"Machine-learning is managing our email junk folders; it is suggesting the next word when we are texting; it is labelling and organising our photo albums; and it makes suggestions on what we should buy next from Amazon or watch next on Netflix."

This is making our lives so much easier and relieves a whole heap of frustration in menial tasks that can be automated. He goes on to discuss how this really useful set of tools could have huge ethical and legal implications for our schools, namely around GDPR (and the storage of student personal data), equity of access to the tools and the propensity for bias in what is likely to be suggested through algorithmic sets. These are too big to delve into here and I would definitely recommend reading Mark's article in full and further reading around ethics in AI. He concludes with a really important statement which I wholeheartedly advocate,

"There needs to be an informed debate about the place of AI in society, and particularly of how it is going to be applied in education."

Let's make that happen! Anyway, back to the AI course...


Chapter 2 begins with a rowboat scenario to explain paths and search/planning tasks that work with AI. This was a brilliantly explained puzzle (where you have to get a fox, chicken and chicken-feed to the other side in a boat) that showed how AI could work in practice. The authors made it really clear that this kind of puzzle could be solved using natural intelligence but if this puzzle were to be multiplied exponentially, this natural methodology would be too complex. However, if one were to employ AI techniques suggested here, you could increase the variables into the millions with little effect on the machine's ability to solve the problem. This is where the real gold comes: increasingly complex problems which the human mind alone would take too long to compute. It reminds me of the quote from Albert Einsten below:


Further into Chapter 2, we are introduced to the Turing machine as a way to understand to compute anything that is computable and then to John McCarthy as the 'Father of AI' who poignantly said:

“The study is to proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.”


This is critical to the understanding of AI and its application to the classroom: every aspect of learning can be broken down to constituent parts so that a machine can perform each step in an automated fashion. Whilst this might not be 100% true (can a computer develop empathy and other human emotions, for instance?), there is some weight to the implications this could have. If a computer has a set of instructions to follow and can be programmed to perform these accurately for a long period of time, the necessity of some human processes may reduce dramatically or disappear altogether. What this means for teachers and/or support staff is yet to be fully thought through. Don't get me wrong, I still believe it takes a teacher but I do think some tasks don't take a teacher!

Games have become the testing ground for much of the AI developments and the concept of game trees with associated nodes and possible outcomes is a low-stakes way to see how AI works in theory. The example given is Tic-Tac-Toe (which I found out is the US name for Noughts and Crosses...a simple Google search for Tic Tac Toe lets you play a cheeky little game in the browser, hence the lateness of this blog post!). The diagram below shows the possibilities from quite a way through the game (from the beginning would be nearly impossible to display!) as what is known as a game tree.


When it comes to games and the associated trees (like the one above), this is fine for simple sets of instructions using what is known as the Minimax algorithm. For more complex game, such as chess, even 10 moves has 2758547353515625 (that’s about 2.7 quadrillion) nodes! They have alternative options and how we can use heuristic evaluation function


So to that end, the idea of games being programmable (to an extent), using algorithms to work smarter and breaking complex ideas down into constituent parts in order to rebuild them all have their place in the education sector. I look forward to Chapter 3 where we look at Real-World AI.

#heuristics #algorithm #AI #game #gamification #Automation #chatbots #machinelearning #ethics #education #GDPR

JOIN THE EDUFUTURIST COMMUNITY

  • email icon
  • Facebook - White Circle
  • Twitter - White Circle
  • Spotify - White Circle
  • apple
  • google pod

© COPYRIGHT EDUFUTURISTS 2020