You can't go online these days without being bombarded by the latest AI breakthroughs. Just when you think you've seen the best, something better even comes up. Like Anthropic who just released Claude 3 Opus, which is now the most powerful large language model in the world, surpassing Google's Gemini Ultra and OpenAI’s ChatGPT-4.
Fun fact: Claude was named after Claude Shannon, a mathematician often referred to as ‘the father of information theory’. He once said “I visualize a time when we will be to robots, what dogs are to humans. And I'm rooting for the machines.” Guess which AI model just hinted at sentient abilities👀
All this AI hype can feel overwhelming and make you wonder if we're heading into a dystopian future ruled by machines. And to be honest with you, I’m tired of it all.
I don't know if this is a good thing. But one thing I know for sure is that these AI wars will go on for a long, long, time. And we can't stop this AI train from getting into education. I'm not sure if we should. I think our best option is to jump on board and steer it in the right direction. I think teachers need to get strategic about guiding students to use AI for more than rote question-answering.
In this article, I'll make a case for why it's crucial to have students using AI assistants as true brainstorming partners - not just polymath fact-dispensers. More importantly, I'll provide extremely practical tips on how to design assignments and activities that unleash AI's potential while keeping human intellect firmly in the driver's seat.
The goal? To nurture a generation of students who can think critically, identify knowledge gaps, and purposefully collaborate with AI. Because let's face it, mastering that man-machine partnership is going to be an essential skill moving forward.
The Matrix Had It Wrong
Remember that mind-bending scene in The Matrix where Neo is uploaded with kung-fu skills in a matter of seconds? As awesome as that capability would be, it's just not how human learning works. True knowledge can't be simply downloaded - it has to be cultivated through active mental work.
And that's exactly why critical thinking and "learning how to learn" are such vital skills in this age of AI and information overload. We're bombarded with a firehose of data and claims from all directions daily. Having the ability to sift through that deluge, identify what's fact vs fiction, and pinpoint gaps in our own understanding is the key to actually building usable knowledge.
So let's look at why explicitly teaching these metacognitive skills needs to be a top priority, using AI as the catalyst rather than a crutch...
1. Separate Fact from Fiction
We all know how easy it is for misinformation and opinions to spread like wildfire online these days. Social media has become humanity's collective stream of consciousness - for better or worse.
But here's the thing - learning to detect garbage is a superpower in this environment. And it's one area where humans still have a critical advantage over AI's pattern-matching abilities. Students who can meaningfully interrogate sources, cross-reference claims, and sniff out flaws in reasoning will be equipped to thrive.
Just think about it - when an AI confidently states "the cause of creative destruction in societies is XYZ", a critically thinking student should be able to:
Identify gaps or assumptions in the AI's knowledge
Compare to expert opinions and authoritative sources
Articulate well-formed queries to further investigate the claim
Ultimately reach their reasoned perspective on the matter
This 2021 study by Stanford research psychologist Sam Wineburg found that students who practiced techniques like fact-checking, evaluating source credibility, and articulating knowledge gaps showed significant improvements in critical reading, reasoning and media literacy skills over the course of a semester.
Without this skill, we risk raising a generation of humans who mindlessly accept what their AI assistants tell them as gospel. And as we've seen time and again, blind trust in technology is a dangerous game.
2. We've All Been There...
Whether trying to ask a question, explain a concept, or just articulate what's rattling around in our brain bucket, we've all experienced the frustration of not being able to translate our thoughts into clear words.
But being able to effectively communicate one's knowledge gaps, theories, and inquiries is crucial for productive learning. It's the difference between "I'm confused" and "I don't fully understand aspect X because of missing piece Y." The latter prompts a conversation, the former, a dead-end.
The process of probing an AI's knowledge explicitly reveals one's own comprehension level. And by practicing how to ask clarifying questions and articulate counter-arguments, students hone invaluable communication muscles.
3. The Buck Stops With You, Kid
At some point, every student needs to shed the training wheels and take ownership of their own learning journey. Spoon-feeding information only gets you so far - real mastery requires taking the initiative to explore, struggle, and synthesize on your own.
And that's exactly why nurturing those self-directed learning muscles is so crucial. The ability to independently identify what you don't know, chart a path to filling those gaps, and persistently cycle through that process is what separates active learners from passengers along for the ride.
Fortunately, AI assistants can be the perfect tool for sparking and guiding that self-driven quest for knowledge if used strategically. Think of them as Virgil to your Dante - illuminating the path when you inevitably find yourself in dark woods of confusion.
Some simple techniques to promote student self-direction:
The Rookie Move: Provide minimal Context on a new topic, just enough for students to get their feet wet researching with an AI. Then have them identify the shallow areas and map out key questions they still need answered through supplemental queries and digging.
Thought Pong: Allow students to volley ideas and drafts back and forth with the AI, using its responses to iteratively refine their thinking. The key is putting them in the driver's seat to steer the direction based on the AI's rapid-fire feedback.
Curiosity Jumpstart: Challenge them to develop a "recursive question" - one that continuously spawns new lines of inquiry the more you explore it, requiring persistent independent research. Few things are more motivating than unanswered questions!
There will always be new unknowns to unravel. The students who can harness tools like Claude to actively navigate their own learning path will be equipped to thrive in our perpetually evolving world.
Enough Theory, Let's Get Our Hands Dirty
Okay, by now you're probably thinking - "Yeah, yeah, I get why developing critical thinking and learning skills is so crucial. But how do I actually have students using AI for that purpose?"
Fair question. It's one thing to grasp the ideals, but pivoting your entire classroom approach to leverage AI in productive ways is the real challenge. That's why I want to get tangible with easy-to-implement techniques.
No more philosophy or vague platitudes about unleashing human potential. We're going into the coding kitchen and cooking up some fresh recipes to get those young minds fired up!
Because students are the biggest supporters of using AI tools...as long as we make the process engaging, thought-provoking, and not just mindless busy work.
1. The AI Prototype-a-Thon
We're all familiar with the dread of writing that first draft - whether it's an essay, story, or code project. Staring at the blank page can induce anxiety and make the whole task seem overwhelming. But what if there was a low-stakes way to rapidly sketch out a rough initial prototype?
The concept is simple - give students a brief on the writing prompt or coding goal, then conscript the AI assistant to generate a basic draft or framework to get the creative juices flowing.
From there, the real magic happens. Students work in teams to critique the AI output, identify flaws or gaps, divide and conquer research to address holes, and iteratively improve the prototype through multiple critique/revision cycles.
Not only does this build crucial collaboration and communication skills, but it also transforms the AI from an endpoint into a true jump-starter for the messy initial ideation phase. Students gain confidence to get past the blank page while practicing meaningful feedback skills.
By leveraging the AI's rapid proto-production and feedback cycles, you can run multiple Prototype-a-Thon sprints in a single class period. Watching bare-bones concepts blossom into cohesive, student-owned creations is a beautiful thing.
The end goal? Having AI shoulder the initial creative burden provides a launchpad for students to then take full ownership in polishing and extending the work to their individualized vision through good old-fashioned critical thinking.
2. The Biased Buddy Showdown
Let's face it - even the most advanced AI models like Claude can have insidious blind spots and biases baked into their training data. While they’re highly capable in many domains, they’re certainly not infallible or objective oracles of truth.
And that inherent fallibility is actually a feature, not a bug, when it comes to developing students' abilities to identify skewed perspectives and dig for counter-arguments.
The Biased Buddy Showdown is an interactive classroom exercise that turns AI bias into a virtue by having students purposefully probe for an assistant's blind spots on particular topics. Here's how it works:
First, propose a controversial or nuanced subject to explore - something without a clear right/wrong stance. Solid options could include historical events, political issues, ethical dilemmas, etc.
Next, have students work in teams to feed different contextual framing into an AI assistant and compare the responses. The key is to probe from multiple angles by tweaking factors like:
Geographic/cultural contexts provided
Stakeholder perspectives emphasized
Moral axioms or ideological lenses
Teams should document how the AI's outputs potentially reflect skewed or incomplete knowledge based on those contextual nudges.
Then, after harvesting the AI's "biased buddies", each team argues why their framed perspective is most well-rounded and logical based on authoritative evidence. Opposing teams get to cross-examine and poke holes, and heated debate is encouraged.
You can raise the stakes by having other teams attempt to break an AI's logic through interrogation. Or run prompts through different language models and award points for poking holes.
The goal here is to turn AI bias into an opportunity for students to actively practice skills like:
Identifying skewed premises and incomplete information
Researching credible counter-evidence
Formulating reasoned arguments and rebuttals
And considering issues through multiple stakeholder lenses
Oh, and one last perk - a few classroom activities spark engagement like a respectful academic throwdown. The holder of "Least Biased Buddy" can bask in bragging rights for days. Every teacher's dream. I know.
3. The Socratic Cyborg
We've all had that experience - asking a follow-up question to an AI response only to get a completely unrelated, nonsensical answer. It can feel like conversing with someone losing their train of thought mid-sentence.
But rather than getting frustrated, why not turn those non-sequiturs into invaluable learning opportunities? I'm talking about consciously putting AI assistants into "Socratic mode" to have students practice the art of disciplined questioning.
Here's how a Socratic Cyborg session could work: Give students a reasonably complex or open-ended prompt to start a dialogue, like analyzing a moral dilemma, interpreting literature/art, or exploring a scientific concept.
From there, the human does the driving while the AI assistant goes full Mr Spock - providing responses based solely on the previous statement without any retained context memory.
The student's job is to:
Identify when the AI's response is incoherent or illogical
Pinpoint what key premises or knowledge is missing
Re-steer the conversation through precise follow-up questions
Basically, they have to practice drilling down the Socratic style to uncover the AI's reasoning path and gaps. Counterintuitive responses are treated as opportunities to re-scope and articulate better prompts.
You can gamify things by scoring points when the AI finally gives a coherent response integrating all necessary context. Or embrace the herky-jerky conversation and deputize students to call out each other's conversation-steering mistakes.
Doing this will help them with navigating ambiguity and fighting through obstacles in the pursuit of deeper understanding. Skills like separating valid from flawed logic, posing clarifying questions, and sticking with an interrogative line of inquiry despite setbacks.
4. The Infinite Curiosity Playground
You know that giddy sensation when you stumble into an entirely new rabbit hole of questions and research paths to explore? That childlike curiosity that blossoms from each new realization of what you don't know?
It's one of the most powerful driving forces for self-directed learning. And relevant AI assistants can be the ultimate curiosity catalysts if used intentionally.
The Infinite Curiosity Playground is a classroom scaffold for sparking those boundless "...but what about?" moments and guiding students to proactively map their own investigation paths.
Here's how it works: Start by having students craft an open-ended, thematic question they're genuinely curious about. The more multifaceted the better - "What causes war?" or "How does creativity work in the brain?"
From there, they enter a cyclical process:
i) Research the topic with an AI assistant, noting where the responses raise more questions than answers.
ii) Clearly articulate the new inquiries and knowledge gaps that emerged.
iii) Prioritize the new threads they want to pull on further.
iv) Repeat the research cycle, allowing that curiosity to rapidly branch out in organic directions.
The teacher's role is to circulate and make sure students are diligently documenting their evolving curiosity pathways and unanswered questions. Periodically they present updates on where their investigations lead.
You can incorporate game elements by awarding points for the degree of topic exploration, number of credible sources cited, creativity of connections made, etc. Gamifying that intrepid detective spirit!
The key is fostering a judgment-free environment for students to bravely pursue their interests without fear of "dumb questions." Celebrate the PI's who get utterly consumed down fascinating new avenues!
The disposition to relentlessly ask "What's next?" is what separates lifelong students from static completers. Why settle for today's understanding when new vistas await?
5. The AI Antithesis Olympics
We've covered some great methods for leveraging AI as a collaborator and curiosity accelerant. But sometimes, the best way to reinforce critical thinking is through productive disagreement and opposition.
That's the spirit behind the AI Antithesis Olympics - a classroom exercise designed to force students into habits of counterargument, scrutiny of premises, and forming substantiated contrarian viewpoints.
Here's how it works: Kick things off by having students wrestle with a complex, subjective prompt related to your current unit of study. Potential examples:
"In the text/data, the author/evidence suggests X. Agree or disagree?"
"New technology Y should be adopted because of Z reasons. Poke holes in this stance."
"What are the most glaring limitations or shortcomings of Theory X?"
Students first work independently or in teams to fully unpack the provided stance using an AI assistant for supplemental context and brainstorming.
The twist? They then trade viewpoints and have to construct a detailed antithesis to argue against the original perspective and underlying reasoning. Basically, playing devil's advocate on steroids.
This forces them to dig into the primary sources, identify flawed assumptions or gaps, and formulate a coherent counter-narrative. The AI can be a tool for moments like:
"OK Claude, analyze this perspective and tell me the weak spots to target."
"What are the strongest counterarguments against this line of reasoning?"
"Help me role-play the opposing side and stress test my counterpoints."
You can place constraints like requiring a minimum number of specific evidence citations or limiting AI use to bounded queries rather than open-ended generation.
The real competitive fun starts when teams engage in structured debates, cross-examinations, and live rebuttals! Gamify things with awards for most effective criticism, killer rejoinders, etc. The smack-talk and showmanship will be unmatched.
The goal is to turn AI from an ambient resource into an active contrarian sparring partner. By assigning the burden of antithesis, students must sharpen skills like scrutinizing logic holes, weighing tradeoffs, being persuasive, and not just passively accepting claims.
The path to truth runs through the crucible of negation. Only once our ideas survive rigorous scrutiny and counterargument can we embrace their validity. Those are mental muscles society desperately needs more of.
In a nutshell
The techniques outlined in this article, from AI prototyping to Socratic interrogations to structured disagreement, are designed to make a paradigm shift. They transform AI from an endpoint into a catalytic starting point for curiosity, analysis, and progressive understanding.
But the strategies are just frameworks. Their true impact emerges from the passion and pedagogical artistry of teachers. It's your guidance that shapes productive AI exchanges, nurturing students to be relentlessly inquisitive, collaborative, and confident life-long learners.
The Value Junction is free for everyone. But you can decide to support my work by sharing it if you find value in this content. Thank you so much!
Bechem, thank you yet again for actionable guidance. I assume [appropriate caveats here] that you encourage your clients to explore and experiment using your approach. In keeping with the theme of the article, it would be interesting to have students who have tried your approach write a case study of their experiences, possibly contrasting the exercise to their conventional learning experience.
Another exercise category might be the King Eurystheus Challenge. Rather than pitting Hercules against the hydra, they could formulate a prompt that could be asked of several AI assistants. I tried this a few weeks ago with the latest versions of Claude, Gemini, ChatGPT, and Perplexity. Gemini and ChatGPT either declined to respond to the prompt or lectured me. After the standardized first prompt, I took the challenge to the fourth level of dialogue with each, keying off their initial response. Interesting. The purpose of this exercise is to help the students uncover bias, evaluate comprehension, clarity/accuracy of response, etc. It will should also teach the students that they shouldn’t unquestionably accept an AI assistant’s responses. By developing a standard prompt, they can submit it to newer revs as they’re introduced.
Mortimer Adler (see The Paideia Proposal) would be proud of your work.
Just focussing on the theory, I love your suggestions on using AI to encourage critical thinking and identifying gaps in their own knowledge and articulation. There’s so much more to come and we have to be ready to make the most of it!