Refocusing Ontario Education on Student Learning and Equity Through Artificial Intelligence

I've spent almost 20 years in public school classrooms fighting for better student learning outcomes, often while facing bureaucracy that pushes back in order to retain a status quo that supports their privilege. I don't have an office hang on to, my classroom is my office and my interests have always aligned with making that learning environment as effective as I can make it.

The pandemic has cast a harsh light on this lack of focus on pedagogy in our education system.  This past year could have been a huge step forward for Ontario education in terms of leveraging technology to produce better learning outcomes, but instead of a Bill Davis style, rational, progressive conservative clean up of an education system steeped in almost two decades of liberal 'vision', we got the Ford circus.  Ontario really deserves better politicians than it gets.

In my time in Ontario classrooms I've seen #edtech evolve at a fantastic rate and I've always kept up with it.  #Onted is a traditionalist organization with many stake holders (unions, boards, ministries, colleges and many other hangers-on too numerous to mention) who are more interested in playing politics in order to justify their role in an increasingly bloated and outdated system.  The pandemic has made it clear to me that most of these groups are focused on doing whatever it takes to keep their office jobs no matter how cruel or harmful to students the plan is.  My union just sent me another email about how we need to start another political fight over EQAO.  That this arrives in a year of historic workplace abuse in the system shows just how tone deaf my union has become.  No one seems to be focused on what matters anymore (student learning outcomes, remember?). 

Dr Sasha Noukhovitch, a fellow CyberTitan coach and colleague, shared an interesting while paper from The Canadian Commission for UNESCO on how artificial intelligence can revolutionize education.  This nuanced look at how AI could provide differentiation and support for all students regardless of their socio-economic situation (assuming we ever make a serious effort to permanently close the digital divide) represents a better understanding of the technology than that shown by the 'robots will take our jobs!' crowd and suggests a pathway toward a future where technology works to provide equity rather than what we're doing with it now.

In a year where everyone likes to talk about equity while doing the exact opposite setting up hugely inequitable pandemic learning schedules, the idea that a an apolitical, rational and student needs focused system could be brought to bear is thrilling.  It's an ongoing frustration that focusing our classrooms on pedagogy feels more and more alien; everyone in Ontario education has lost the plot and left it to exhausted and under-supported classroom teachers to make their inequitable planning work.

Artificial Intelligence offers the kind of individual support specific to student needs that the system has always struggled to provide.  I've been dreaming about it for ten years.  Our low-resolution bureaucracy does an adequate job of managing a mythically average student but doesn't like to treat students like people because that costs money.  AI could do a lot to address that inability to address equity, but rather than explore this emerging technology you can bet the privileged/political stake holders will do all they can to block it in order to maintain their status quo benefits.

This is about the UK but
the conservative playbook
looks the same everywhere.

The second article from The Guardian about British schools offers some worrying details about how behind the curve they are in terms of technology adoption (lots of schools don't have wifi yet?  C'mon UKed!).  It also suggests a way to improve student learning outcomes that has become apparent from asynchronous online learning: "One way to tackle the achievement gap is surely in-school lessons followed by more personalised online learning, either at home or in after-school clubs."  Of course, in Ontario we rush to apply technology to force synchronous learning (recreating the inequities of the classroom) for political ends while further crushing students whose families can't provide the infrastructure.

Combine the concept of immanent personalized virtual learning AIs that will tirelessly support students right where they need it and the idea that school can happen both in class synchronously and out of class virtually and at the student's own pace and you have a recipe for a quality of pedagogy that we simply can't produce in our status-quo, politically charged bureaucracy intent on retaining all the infrastructure (schools, board offices, union offices, educational hangers-on...) and the jobs needed to run it.  A leaner burning Ontario education system focused on student learning might have a similar number of people working in it but almost all of them would be actually involved in teaching.

The thought of a rational, politics free AI focused entirely on maximizing learning outcomes has me dreaming of an education system free of messy human politics and the self-serving political organizations that feed off it.  Decisions would be data driven, transparent and then held to accountability through more transparent data collection that would be ongoing and everywhere rather than centred in a questionable and expensive organization run by a failed politician.

I'm in my final decade of teaching and I've lost faith in my union and doubt the intentions of educational management all the way through the system.  The 'support' organizations that also feed off the education system seem to have completely lost the plot in the political haze of education in 2021 Ontario.  Spending my final years in the system making student supported AI learning tools a reality and watching them burn the status quo to the ground would be a satisfying conclusion  to a career spent focused on student learning.  I've long hoped to leave the system in better shape than I found it.  I think the route to that goal is through adapting emerging artificial intelligence and other digital learning tools through a ruthlessly pedagogical focus.  If that burns our bloated bureaucracy to the ground in the process then I'll have achieved my goal of a more equitable and effective public education system that serves student needs first.

WANT TO DIG INTO AI IN EDUCATION FURTHER?

Read that UNESCO white paper on AI in education if you haven't yet, it's worth a read.

ISTE also has primers on AI and how it will change education here.

The Bar For Being Human Just Got Raised Again

As artificial intelligence continues to improve through both hardware and software engineering advances it'll get to the point where it starts to overtake low functioning human beings.  Then it'll overtake average human beings.  At some point it'll overtake the smartest human beings on the planet.  Ray Kurzweil called this point the technological singularity.  It's the point where we won't know why machine intelligence is doing what it's doing any more than an ant would understand why the humans are having a picnic.

You can read up on GPT-3, the latest step forward in A.I. here, or here, but I'm curious about it from an educational point of view.  GPT-3 is the latest iteration of OpenAI's research into text prediction machine intelligence.  Version three isn't that architecturally different from GPT-2, but it's much, much bigger, by many orders of magnitude.  This brute force approach allows it to adapt and respond with such high fidelity that it surprises people.

What does this mean in education?  GPT-3 based online systems are going to start appearing in the next year.  These systems will take a few suggestions from a human user and create text outputs that will stress a Turing test in terms of how well they are put together and what is being said.  With sufficient training and some smart engineering around focusing inputs, GPT-3 based online systems will write an accurate, original essay on any subject.  It could be used to answer any questions in any subject or formulate text responses even in abstract areas like poetry .  It'll also translate better than anything we've seen so far.  It's GPT-3's agile Swiss-Army-knife effectiveness that will see it falling into student hands sooner rather than later.  Which students?  The ones it already sounds like:

"GPT-3’s ability to dazzle with prose and poetry that sounds entirely natural, even erudite or lyrical, is less surprising. It’s a parlor trick that GPT-2 already performed, though GPT-3 is juiced with more TPU-thirsty parameters to enhance its stylistic abstractions and semantic associations. As with their great-grandmother ELIZA, both benefit from our reliance on simple heuristics for speakers’ cognitive abilities, such as artful and sonorous speech rhythms. Like the bullshitter who gets past their first interview by regurgitating impressive-sounding phrases from the memoir of the CEO, GPT-3 spins some pretty good bullshit."

http://dailynous.com/2020/07/30/philosophers-gpt-3

Dig up some GPT-3 output online and you'll see it uses the fact that it has figured out human speech patterns to smoothly say very little; it's like listening to a slick salesman  This complex machine learning formula is the perfect tool for weak students answering rote, repetitive school assignments, because both those students and the school system they're responding too are so low functioning that this rudimentary A.I. can do the job better (and in less than a second).

"As AI researcher Julian Togelius put it: “GPT-3 often performs like a clever student who hasn’t done their reading, trying to bullshit their way through an exam. Some well-known facts, some half-truths, and some straight lies, strung together in what first looks like a smooth narrative.” (Though as many have pointed out: clever students who know how to bullshit go far in this world because people don’t always scrutinize what they’re saying.)"

https://www.theverge.com/21346343/gpt-3-explainer-openai-examples-errors-agi-potential

So, the bar for human expectation has just moved again.  If you're operating as a teacher or student at the sharp end of human achievement, this is well beneath you, but if you like to trot out the same old material year after year, don't bother assessing process and don't really pay much attention to student work you do mark, this'll fool you.  For a student looking to get something for nothing this is a dream come true.

"GPT-3 would never kill jobs skilled developers. Instead its a wake-up call for cargo coders and developers. It’ll urge them to buckle up and upskill to ensure they’re up for solving complex computer programming problems." (cargo coders are weak programmers who copy and paste code rather than generating it themselves - they're like many students)

https://towardsdatascience.com/openais-gpt-3-the-end-of-cargo-cult-programmers-23102f70f855

The obvious answer to this is to assess process, since a student attempting to hand in work this way would have none.  Of more interest from a pedagogically standpoint is how we should integrate this evolving technology into our learning processes.  OpenAI isn't doing this in an evil attempt to create an entire generation of illiterate children, they're doing it to create A.I. that assists and supports human endeavour and raises it to a higher level.

The last year I was teaching English I had my 3Us try and beat Turnitin.com.  The standard usage seemed to be an 'aha, I caught you plagiarizing!" punitive response after a minimally reviewed writing process, all done behind a curtain.  By turning that all around  and giving students transparent access to this punitive tool, I had students come to the realization that they could beat turnitin's plagiarism check, but it took so much work to do it effectively that it was easier and more functional to just write the damned thing yourself.  Instead of using tech to catch cheating or banning it outright, we used it to test limitations and validate a meaningful writing process.

I imagine education's first response to GPT-3 driven plagiarism tools will be to try and ban them, but as usual that's backwards.  A.I. supported human intelligence isn't being developed for us to do less, but to enable us to do more.  From that point of view, an A.I. supported writing process should move rubric expectations for everyone upwards.  What used to meet expectations should now fail to meet expectations.  A digitally supported writer should already be leveraging tools to mitigate grammar and spelling errors, and teachers should be teaching effective use of these tools.  Where 5-10 grammar errors in a paper might have gotten you a level three/meets expectation before, there should now be none because digital supports should be integrated, proficiency in them expected and output from them meeting raised expectations.  With that technical work supported, student writers should be focusing on developing rhetoric, continuity of thought, voice and style.

The same goes for A.I. supported writing as we enter the Twenty-Twenties.  We should be evolving writing processes to include A.I. editorial review, A.I. supported enhanced research and maybe even A.I. driven originality of thinking.  Can you imagine a Turing test as a part of writing process that tells a student that their writing isn't as human as a GPT-3 piece?  That's using A.I. to raise the bar.  Can you imagine what student writing might look like if advanced word prediction A.I.s like GPT-3 were integrated into student writing processes?  We all need to be thinking about that, now.  It's what literacy is going to look like in the next decade.

Beyond writing you're going to see GPT-3 driven online tools rock rote, standardized (lazy) learning.  Like your worksheets?  A student will be able to scan a worksheet and receive accurate, textually correct responses instantly, to any question, in any subject.  If you're using the same old assignments over and over, the A.I. will find that and use previous examples to produce even more complex and relevant answers.  If you assess student learning through limited, repeated single-assessment final products, you just got punked.

The irony is the teachers who struggle most with this new threshold of human expectation are also the ones who will use it to mark student work.  In those teach-like-it's-1960 classes, A.I. written papers will be handed in by students and then marked by A.I. markers - no humans will have played a part in any of that 'learning'.





OTHER READING

https://www.wired.co.uk/article/gpt-3-openai-examples  "the world’s most impressive AI. Humans are being given limited use – for now – to make sure things don’t go wrong"

https://lambdalabs.com/blog/demystifying-gpt-3/
A technical analysis of GPT-3:  OpenAI recently published GPT-3, the largest language model ever trained. GPT-3 has 175 billion parameters and would require 355 years and $4,600,000 to train

Tech Cleaning Up Its Own Mess: how to fix misinformation in digital media

I'm having one of those intersectional moments where my recent work in AI, coding and cybersecurity have me thinking about ways we can fix the worst parts of our digital adolescence.  Media like the tweet below are wearing everyone down, but I think this is a digital media problem that digital media can help resolve:

Together, we can reignite the heart & spirit of our youth.

#ONpoli pic.twitter.com/DRxUehxYMB

— Stephen Lecce (@Sflecce) February 15, 2020

In this case an elected official is claiming to support children with special needs while at the same time doing the doing the opposite behind the scenes, even going so far as to ignore signed contracts and cancelling support.  As I watched this misinformation I wondered why the digital system delivering it (Twitter in this case), couldn't include links and information to clarify what I'm watching.  Doing so would help users understand when they are being misled.  Can you imagine a digital media ecosystem that actually encourages truth and accuracy instead of what we have now?

From a data management point of view, rhetoric and political spin should bump up against a scientific analysis of fact based initially on volume of data.  Facts tend to have more data behind them (proving things takes time and information).  Attacking this as a big-data computer science project, statements made by politicians could be corroborated by connecting to supporting digital information in real time.  I dream of the day when I'm watching a politician's speech live online on any browser (this should be baked into every browser) while seeing an AI driven analytical tool that is leveraging the digital sea of information we live in to validate what is being said.

This information enrichment would do two things.  Firstly, it would create a truth-tendency over time metric that would allow voters to more accurately assess the accuracy of what politicians, news outlets and even each other are saying - a kind of digital reputation.  Secondly, having an impartial analysis of social activity in real time would mitigate and highlight fake news and help social media to resolve its terrible handling of misinformation.

There are layers and layers to digital misinformation.  As we've moved from lower bandwidth mediums like text through still images to video, misinformation is keeping up, often under the guise of marketing.  You can't trust anything you see online these days:

It's a new form of media literacy that most people are unaware of.  There are plugins attempting to battle photoshopped images and videos that should help stem that tide of misinformation.  Movement on this is fast because parsing image and video data is a more mathematically biased problem, but intentional misinformation either created or shared is also something machine learning systems can get better and better at identifying as they learn the peculiarities of why humans lie to each other.

In the case of something like Vaughan Working Families, a fake organization designed to spread misinformation by wealthy government supporters, the misinformation was fairly easily identifiable by looking into the group's history (there is none).  That lack of data is a great starting point in training an AI big data analysis system in live response to misinformation - the truth always weighs more because of the evidence needed to support it.

We do IBM Watson chatbot coding in my grade 10 computer engineering class, and it is interesting to watch how the AI core picks up information and learns it.  As it collects more and more information, and supported by students teaching it parameters, it very quickly picks up the gist of even complex, non-linear information.  Based on that experience, I suspect a browser overlay that offers a pop up of accurate, related information in real time is now possible.

In software you have the front end that faces the user and the back end that does the heavy lifting with data.  In the cloud-based world we live in, with people sharing massive amounts of data online, an unbiased, ungameable, transparent AI driven fake news overlay would go miles in restoring the terrible history Facebook, Google, Microsoft and the rest have in interfering with democracy.  This shouldn't be something squirrelled away and only available to journalists.  It should be a technical requirement for any browser.

With that unblinking eye watching the dodgy humans, not only would politicians be held to a higher standard, but so would everyone.  Those quiet types who happily retweet and share false information are complicit in this information virus.  If your Twitter account ends up with a red 17% accuracy tag because you regularly create and share misinformation, then I'd hope it results in less people being interested in following you, though I don't personally have a lot of faith in people to do even that.  Left to our own devices, or worse, chasing the money, we've made a mess or things by letting digital conglomerates disrupt institutions that took years to evolve into pillars of civil society.  It's time to demand that they use the same technologies they are leveraging now to fix it.

We're obviously either too lazy and/or self interested to make a point of fact checking our social media use.  If we're all on there sharing information, we should all make our best effort in sharing it accurately.  This could help make that happen.  It would also go a long way toward preventing the the cyber-crime epidemic we live in which thrives on this kind of hyperbole and irrational response.

There have been some attempts by charitable organizations and students to create online fact checkers, but the browser creators (Google, Microsoft, Firefox, etc), and social media giants (Facebook, Twitter, etc) don't seem to be the ones doing it, even though they've gotten rich from this misinformation and damaged our ability to govern ourselves as a result.  Law can't keep up with our technological adolescence and the data avalanche it has produced, but the technology itself is more than able.

We increasingly depend on people, often amateurs with little or no funding, to do our online fact checking, but the sheer volume of information, especially when driven by automated processes like bots, makes that unscalable.  This is something that professional  journalists used to do (at least I hope they used to do it, because not many are doing it now).  However, the financial pressure on those institutions due to digital disruption means they are now more than happy to take inaccurate and misleading information and share it if it makes them somewhat relevant again.  The only way to address this situation is by leveraging the same technology that caused it in the first place.

What do you say tech billionaires?  Could we redesign our digital media browsing so it encourages accurate information rather than making it irrelevant?  You might have to actually put some financial support into this since you've effectively dismantled many of the systems that used to protect the public from it.