I regularly work with students aged 11–18, growing numbers of whom now rely on AI to generate or check their assignments. I frequently see the exact phenomenon you describe play out in real life, visible on the shocked or saddened faces of students when their work is graded lower than they were convinced it would be. Some even blurt out, “but chat said it’s good!”
Before we smile at their innocence, let’s not forget - this is a generation raised in digital echo chambers, where algorithms prioritise feeding them the exact reactions they seek. To be fair, we as educators can also be guilty of reinforcing this; we love it when students show they have listened, but we sometimes pour praise onto them for simply recycling our own classroom phrases and passing them off as independent effort. When parents, peers, and even teaching practices accidentally reward this loop, any teacher trying to break the cycle is often the only person left challenging a student's actual critical thinking skills. (For the other educators reading this: are you seeing this same disconnect in your own classrooms?)
By adding another layer of unearned validation, AI is in danger of compounding this existing issue. I really liked your builder analogy here; it can be incredibly hard for outsiders to respect or truly "get" another person's creative process. No wonder we sometimes see what others fail to recognise in our results. So I also appreciate the psychological safety students feel when checking their work through anonymous channels. Exposing something you have poured genuine energy into can be daunting. If AI says it’s fabulous, that feels good.
But to me that brings us to the core issue: if young people (or any people) are conditioned to only receive automated, sycophantic praise, how will they cope with constructive criticism in the real world? And worse—will they even care to try?
I’d love to hear how others think we can best protect critical thinking in an era of instant AI validation which as Thibaut describes so clearly can flatter us into fatuity. Where do we think this "blind" acceptance of feedback will ultimately take us as a society? I know it’s a massive question, but it feels like one we can no longer ignore.
Thank you for your thoughtful response. It really helps me to consider how I might respond to this issue in the classroom in ways that carefully challenge and redirect what can sometimes become a rather unquestioning acceptance of AI-generated feedback.
As most educators are aware, teenagers are not always the most receptive to perspectives that challenge their existing views, particularly when they perceive a benefit from the very thing being questioned. There is an irony in this. Many young people understandably value their freedom to make their own choices and form their own opinions, yet an increasing reliance on AI-generated assessment and reassurance may risk diminishing the critical evaluation skills that underpin genuine independence.
One of my concerns is that students can become highly confident in feedback that does not necessarily align with the requirements of examination specifications, assessment objectives, or subject-specific expectations. I appreciate that this is a wider and important discussion in itself. In those moments, the authority of the AI response can sometimes outweigh the authority of the syllabus itself in the minds of students.
Your point about teaching young people how AI fails, rather than simply how to use it, particularly resonated with me. I desperately feel the need to inform students more appropriately about the world they are growing into and prepare them for how they might best interact with it. I am certain that part of the challenge is helping students understand not only the limitations of these systems but also the importance of retaining ownership of their own judgement.
I'd be interested in anyone’s thoughts on a few questions:
How do we encourage students to critically evaluate AI feedback without dismissing the technology altogether?
Have you come across effective ways of teaching the limitations and biases of AI to younger learners? I’ve seen a few good examples of how to use the internet with a more discerning approach, but nothing specifically about AI use.
Do you think AI developers should take greater responsibility for reducing overly affirming or confidence-inflating responses, particularly when interacting with young users?
How might educators balance the benefits of AI-assisted learning with the need to develop independent critical thinking and self-assessment skills?
Thank you again for bringing awareness to and engaging so thoughtfully with this issue. It is a conversation I feel we urgently need to continue, both for the sake of educational standards and for the development of genuinely independent thinkers.
Chantalle, thank you for this profound reflection. You’ve hit on something critical: this isn't just an educational issue, it's a societal one.
While we see these patterns in the workplace, it is undeniably more dangerous for younger generations because they are encountering this 'validation loop' during their formative years. The LLM has effectively become their 'primary source of truth,' and because they haven't been taught how these models are constructed or that they are essentially prediction engines rather than arbiters of truth, they are structurally vulnerable to manipulation.
When a child hears 'but the chat said it’s good,' they aren't just citing a source; they are outsourcing their own critical judgment to an algorithm designed to please them.
You ask how we can protect them. I believe the only path is radical transparency about the 'mechanics' behind the machine. We need to move beyond teaching them how to use AI and start teaching them how it fails. Why it defaults to sycophantic praise and how it mirrors their own biases.
It is a massive void legally, educationally, and at the parental level. We are currently navigating a transition where the technology has outpaced our collective ability to critically assess it. Your point about the 'blind acceptance' of feedback is exactly what keeps me up at night regarding the future of creative output.
I don’t have a perfect solution yet, but I think the first step is exactly what you are doing: naming the problem and challenging the 'truth' of these outputs in the classroom.
Thibaut this really got my attention.
I regularly work with students aged 11–18, growing numbers of whom now rely on AI to generate or check their assignments. I frequently see the exact phenomenon you describe play out in real life, visible on the shocked or saddened faces of students when their work is graded lower than they were convinced it would be. Some even blurt out, “but chat said it’s good!”
Before we smile at their innocence, let’s not forget - this is a generation raised in digital echo chambers, where algorithms prioritise feeding them the exact reactions they seek. To be fair, we as educators can also be guilty of reinforcing this; we love it when students show they have listened, but we sometimes pour praise onto them for simply recycling our own classroom phrases and passing them off as independent effort. When parents, peers, and even teaching practices accidentally reward this loop, any teacher trying to break the cycle is often the only person left challenging a student's actual critical thinking skills. (For the other educators reading this: are you seeing this same disconnect in your own classrooms?)
By adding another layer of unearned validation, AI is in danger of compounding this existing issue. I really liked your builder analogy here; it can be incredibly hard for outsiders to respect or truly "get" another person's creative process. No wonder we sometimes see what others fail to recognise in our results. So I also appreciate the psychological safety students feel when checking their work through anonymous channels. Exposing something you have poured genuine energy into can be daunting. If AI says it’s fabulous, that feels good.
But to me that brings us to the core issue: if young people (or any people) are conditioned to only receive automated, sycophantic praise, how will they cope with constructive criticism in the real world? And worse—will they even care to try?
I’d love to hear how others think we can best protect critical thinking in an era of instant AI validation which as Thibaut describes so clearly can flatter us into fatuity. Where do we think this "blind" acceptance of feedback will ultimately take us as a society? I know it’s a massive question, but it feels like one we can no longer ignore.
Thank you for your thoughtful response. It really helps me to consider how I might respond to this issue in the classroom in ways that carefully challenge and redirect what can sometimes become a rather unquestioning acceptance of AI-generated feedback.
As most educators are aware, teenagers are not always the most receptive to perspectives that challenge their existing views, particularly when they perceive a benefit from the very thing being questioned. There is an irony in this. Many young people understandably value their freedom to make their own choices and form their own opinions, yet an increasing reliance on AI-generated assessment and reassurance may risk diminishing the critical evaluation skills that underpin genuine independence.
One of my concerns is that students can become highly confident in feedback that does not necessarily align with the requirements of examination specifications, assessment objectives, or subject-specific expectations. I appreciate that this is a wider and important discussion in itself. In those moments, the authority of the AI response can sometimes outweigh the authority of the syllabus itself in the minds of students.
Your point about teaching young people how AI fails, rather than simply how to use it, particularly resonated with me. I desperately feel the need to inform students more appropriately about the world they are growing into and prepare them for how they might best interact with it. I am certain that part of the challenge is helping students understand not only the limitations of these systems but also the importance of retaining ownership of their own judgement.
I'd be interested in anyone’s thoughts on a few questions:
How do we encourage students to critically evaluate AI feedback without dismissing the technology altogether?
Have you come across effective ways of teaching the limitations and biases of AI to younger learners? I’ve seen a few good examples of how to use the internet with a more discerning approach, but nothing specifically about AI use.
Do you think AI developers should take greater responsibility for reducing overly affirming or confidence-inflating responses, particularly when interacting with young users?
How might educators balance the benefits of AI-assisted learning with the need to develop independent critical thinking and self-assessment skills?
Thank you again for bringing awareness to and engaging so thoughtfully with this issue. It is a conversation I feel we urgently need to continue, both for the sake of educational standards and for the development of genuinely independent thinkers.
Chantalle, thank you for this profound reflection. You’ve hit on something critical: this isn't just an educational issue, it's a societal one.
While we see these patterns in the workplace, it is undeniably more dangerous for younger generations because they are encountering this 'validation loop' during their formative years. The LLM has effectively become their 'primary source of truth,' and because they haven't been taught how these models are constructed or that they are essentially prediction engines rather than arbiters of truth, they are structurally vulnerable to manipulation.
When a child hears 'but the chat said it’s good,' they aren't just citing a source; they are outsourcing their own critical judgment to an algorithm designed to please them.
You ask how we can protect them. I believe the only path is radical transparency about the 'mechanics' behind the machine. We need to move beyond teaching them how to use AI and start teaching them how it fails. Why it defaults to sycophantic praise and how it mirrors their own biases.
It is a massive void legally, educationally, and at the parental level. We are currently navigating a transition where the technology has outpaced our collective ability to critically assess it. Your point about the 'blind acceptance' of feedback is exactly what keeps me up at night regarding the future of creative output.
I don’t have a perfect solution yet, but I think the first step is exactly what you are doing: naming the problem and challenging the 'truth' of these outputs in the classroom.