Will AI replace human feedback?

Yes and no.

The advancements of AI in the last year are nothing short of extraordinary and these advancements will only compound at exponential rates.  

In a recently released working paper by Arran Hamilton, Dylan Wiliam and John Hattie called “The Future of AI in Education: 13 things we can do to minimize the damage” (2023), Hamilton et al. present 4 somewhat dystopian scenarios that the development and expansion of AI could lead to:

1. AI is banned: Governments unite to put a brake on progress.
2. AI and humans work side-by-side aka. fake work: AI is capable of doing most jobs, so governments legislate to keep people in the workforce to maintain a sense of purpose in society.
3. Transhumanism: Where we upgrade our brains (it’s easy to picture Neo learning Kung Fu in “The Matrix” here).
4. Universal Basic Income: Where we decouple from the economy and let the machines run the world, while humans pursue art, sport and leisure activities (we’ve seen this before in Disney/Pixar’s “Wall-E”).

Now if we take a step back and assess these 4 scenarios, they all look somewhat bleak.

1. = Restricting progress.
2. = False purpose.
3. = Opportunity for abuse of power.
4. = A potential utopia or dystopia? Dependent on your proclivities in life.

I first met Dr Arran Hamilton in March 2023 to show him Vibbl. What I expected to be a 30 minute demo conversation ended up being a fascinating conversation about AI that lasted for over two hours. At the time, I thought Arran’s views on AI were alarmist and I struggled to picture the future scenarios happening as quickly as he thought. The truth is Arran was just far more informed on AI than I was at the time and frankly, I was wrong. My ignorance was blocking my ability to see clearly as what Arran was saying was such an enormous change. I was at the denial phase of the five stages of grief towards the world I now know.

These scenarios may also seem alarmist to you, particularly if you have only just started to look into AI. It adds no comfort that some of the most prominent voices in the AI space predict similar and sometimes more extreme outcomes (Sam Altman, Geoffrey Hinton, Mo Gawdat and many more). However, it is important to note that not all AI experts agree, some notable voices like Prof Yann LeCun does not see AI as dangerous, but still believes it will surpass human intelligence and replace many jobs. 

I personally think that all 4 are very likely outcomes, with the potential for all scenarios to be lived in some capacity.

I predict we may see certain countries with more totalitarian politics restricting AI to the government, others where fake work is legislated, universal basic income – perhaps in richer countries and transhumanism being experimented with for the super wealthy.

So, how will this all affect education?

Well, Hamilton et al. provide a range of scenarios that prompt how AI may impact education, such as through creating challenges for authenticity in assessment, marking homework, creating resources for teaching and many other ways AI could reduce teacher workload, or possibly replace teacher workload.

We are already seeing many of these risks and opportunities happening with the current version of Chat GPT.

I was lucky enough to share the stage with Dan Fitzpatrick aka. “The AI Educator” earlier this year as part of an expert panel on AI in education at Newcastle University. Dan was exceptional and gave a great account of how AI could affect education, but what really struck me was his schedule. He literally had triple the tour dates of Beyoncé! I had never seen such a packed schedule and I used to teach!

The point is, school leaders all over the world are scrambling to understand the implications of AI; it’s a hot topic right now. Dan is in great demand right now because he is both excellent and riding a wave of intrigue into AI.

The result of this intrigue is that we will see more and more schools utilising AI in the coming months and years, this is inevitable. But, will AI replace teacher feedback and replace our teaching jobs?

My opinion diverges a bit with the general AI consensus.

I think the education profession is probably one of the safest professions and the way schools operate is likely to stay broadly similar. Here’s why:

1. School is more than just education. It is also childcare.

The Prime Minister who stands in front of Number 10 and announces that human teachers will be replaced by AI because it is more intelligent and can teach your children better is a fictional character akin to something you would see in Black Mirror.

(Unless our national leadership is taken over by AI of course, which may be more politically palatable and likely …)

There are already AI tutoring schools which purport to considerably increase learning in maths and English and involve children sitting at a computer and interacting with an AI tutor. Although this has many use cases, it is not at a place to replace schools and teachers. It is an option for alternative schooling at best for counter-cultural parents and as a supportive tool for traditional education.

Even if we had AI robots that could deliver life like teaching and contain impulsive Year 9’s in a physical building, I think parents would ALWAYS be more comfortable with humans watching their children than robots. This is the key…

Even if we no longer had jobs – had no requirement to learn – most parents would still want to send their children off to a “place” to learn “stuff”. Currently, school provides childcare for parents who are fuelling our human economy, but if you reflect back to bygone years, the echelons of society who had the means to thrive without working would typically send children off to a school of sorts. No matter how realistic and likable robots may be, I can’t see human nature shifting with the future of how our children learn at stake.

 

2. We are all social animals

It is hardwired into us to seek community and even if AI could teach better than humans, there is value to be derived from children learning to interact with adults and peers.

I reflect here on the analogy of a famous vegan food manufacturer called ‘Beyond Meat’. They are truly an incredibly innovative company who are now facing huge challenges as consumers start to shift away from meat replacement foods. They once invented a vegan burger that would bleed as it was cut.

Consider this for a moment, the millions of dollars that must have been spent in research and development to create this engineering marvel. This year, sales have slumped in fake burgers with the boss of Quorn blaming “bleeding burgers” as one of the reasons for the shift in consumer attitudes.

Just because something CAN be done, doesn’t mean it SHOULD be done.

No matter how impressive, AI intelligence and robotics may become, it will never be human. Humans are tribal and AI will always be the other.

I’m aware that similar arguments were made in the past about how customers would not shift away from human tills, from buying cars in person and from visiting the high street to shop online.

Time has proven that we prefer convenience and speed over interaction with lots of processes. But the process of education is different, there are multiple stakeholders involved in the decision-making process and also several motivational drivers for wanting children to go to a school e.g., “to get qualifications”, “to learn”, “to make friends”, “to allow [the parents] to work” –  the process is not transactional like buying a ladder from Amazon. It is complex, nuanced, messy – the kind of process which requires humanity.

Human-led education is here to stay.

3. The best feedback, is recursive feedback

Feedback should not be a one-way conversation. It should be a circular process like the economy we aspire towards – or the fundamental teachings of Mufasa in “The Lion King”.

When teachers read through student work the insights they gain form the basis of the next lesson they will plan, how they will address the needs of different students. Feedback is a part of the pedagogical process, it is not a separate bolt on that can be optimised and standalone. “Formative assessment cannot be fully understood except within the context of a theory of pedagogy.” (Black & William, 2018).

Similarly, feedback should push students forward. It should take into account prior learning, where am I going? How a student is progressing, how am I going? And where the student needs to get to. Where to next? (Hattie and Timperley, 2007).

Much of this could realistically be replaced by language learning models and AI, but I don’t think it will be. Mainly due to my first two arguments. If we accept that humans are leading education and that education is still happening in a physical setting, there would be no point in replacing the feedback with AI as it would weaken the capability of the teacher.

We could have a hybrid model where AI provides analysis for teachers to act on, so that teachers actually don’t need to read any work, but that would be like asking a landscape artist to paint a masterpiece using Google images for inspiration. I think these tools will thrive, but will support practice, rather than replace practice.

 

4. In a future of AI, workload may be the new rich

Our society has hit a workload epidemic, it is now impossible in many parts of the country to afford a house without two, or even more incomes. Email and social media has meant that work is always coming into our inbox and we are always comparing ourselves with others who have more, have done more and are more happy…allegedly.

Now imagine all of that stopped, and suddenly AI had removed almost all purpose from your life. I argue that teaching is unlikely to be replaced, but I believe many other jobs will be replaced by AI. Doctors, lawyers, accountants and many more. Transactional services where being right and a quick resolution are more important than the experience.

You no longer have jobs to do around the house, you don’t need to work, you don’t need income. What do you do with your time? It seems plausible in this scenario that there will be select “vocations” where AI cannot easily replace humans and these, I suspect will become high demand and elitist. Teaching, I believe will be one of them. Humans crave purpose, and in a world with no purpose the lucky few that get to be working on giving students feedback may just be the influencers of the future.

I guess the rest of us can just watch Netflix and chill.

Refences:

• Black, P., & Wiliam, D. (2018). Classroom assessment and pedagogy. Assessment in Education: Principles, Policy & Practice, 25, 551 – 575.
• Hamilton, A. et al. (2023). The Future of AI in Education: 13 things we can do to minimize the damage.
• Hattie, J.A.C., & Timperley, H (2007). The power of feedback. Review of Educational Research, 77(1), 81-112