Teaching "AI and Jobs" to Architect mode (ages 15–18)

Hook & Warm-Up

Architect mode has no mascot and treats students as near-adults making real decisions about their own future — several of them are choosing a college major or a first career direction within the next year or two. This lesson lands differently for them than for any younger age band: it's not abstract.

Say: "AI isn't just replacing jobs — it's reshaping what skills matter and how we work. Let's examine the future of careers in an AI-powered economy." Then ask directly: "How many of you have already thought about whether the field you're interested in is 'safe' from AI? What did you conclude, and what did you actually base that on?" Let a few students answer — most answers at this stage are based on vibes, a headline, or a single anecdote rather than any real framework. Name that gap honestly: "By the end of today, you should have an actual framework for answering that question, not just a feeling about it."

It's worth being direct about the stakes here in a way you wouldn't be with a younger class: "This isn't hypothetical for you the way it might be for a younger sibling. Some of you will be applying for your first job in the next few years, in a labor market that's already being shaped by this technology. The goal today isn't to scare you — it's to make sure you're making decisions with real information."

One framing that tends to land well with this age group: this lesson isn't really about predicting the future — nobody, including professional economists, can predict exactly how any specific field will look in ten years. It's about building a habit of analysis you can reapply for the rest of your working life, because the technology, and the jobs it touches, will keep changing well past whatever specific facts this lesson covers today. The framework matters more than any individual example.

If a student asks whether you personally think their intended career is "safe," resist giving a confident yes-or-no answer even if you have a strong opinion — model the exact analytical habit this lesson is trying to build instead: "I don't know for certain, but let's break the job into its tasks and reason through it together." That response is more useful to them than any reassurance, and it's more honest than pretending anyone has a reliable ten-year forecast for a fast-moving technology.

Main Activity

Have students work through the three scenes independently first, then run the debrief as a structured discussion focused less on "which jobs are which" and more on "what determines the answer, and what should I personally do about it."

Scene 1 — Jobs Changing. The four examples — doctor, lawyer, teacher, designer — are each a case study in partial automation. Frame it explicitly using the task-based lens: a doctor's job includes diagnostic pattern-recognition (increasingly AI-assisted) and patient relationship management, informed consent, and complex judgment under uncertainty (not). A lawyer's job includes document review and legal research (increasingly automatable) and oral advocacy, negotiation, and client counsel (not). A teacher's job is overwhelmingly relational and mentorship-based, which is why it's the lesson's clearest example of resistance to automation. A designer's job includes rapid option generation (now AI-assisted) and the higher-order judgment of which option actually serves the audience and the brief (not). The throughline: automation risk attaches to specific tasks, not job titles, and every real job is a mix.

Scene 2 — New Jobs. Prompt engineering, AI ethics officer roles, and AI training/data labeling are all real, current job categories — not speculative future jobs. It's worth being precise about each: prompt engineering is the practice of designing and iterating on instructions to reliably get useful output from a generative AI system, valuable because these systems' outputs are sensitive to how a task is specified; AI ethics or "responsible AI" roles exist because deploying an AI system carries real risk of harm (bias, safety failures, misuse) that organizations are increasingly expected — by regulators, customers, or their own risk management — to actively manage; and AI training/data-labeling work is the large, often underappreciated human labor that most modern AI systems depend on to learn what a "good" output looks like. Worth naming honestly: some of these roles, particularly at the more routine end of data labeling, are themselves lower-wage and outsourced work, and are not guaranteed to be durable long-term careers just because they're "AI jobs" — the field being new doesn't make every job within it automatically secure.

Scene 3 — Future Skills. This is where the lesson's most substantive claims live, and where the quiz's vocabulary comes from. Introduce Autor's task-based framework directly: rather than asking "will this job be automated," economists studying labor and technology (notably David Autor's body of work) argue for decomposing a job into its component tasks and evaluating automation potential task by task. This reframes the question from a binary (safe/unsafe) into an analysis (which parts, how much, how soon). Introduce "entry-level compression" as a live, debated concern rather than a settled finding: current discussion in labor economics and among employers considers whether AI tools performing tasks traditionally assigned to junior workers — first drafts, preliminary research, routine coding — could reduce the number of true entry-level openings even in growing fields, which matters enormously to a class about to enter that exact labor market. Finally, introduce the skill premium shift: the argument that the economic value of "complementary" skills — those that become more valuable specifically when combined with AI, like knowing how to direct, evaluate, and correct an AI system's output — is rising relative to skills that AI can substitute for outright.

It's worth pausing to distinguish this from a simpler, weaker claim: "just learn to use AI tools and you'll be fine." A complementary skill isn't merely knowing which button to click — it is the judgment to know when an AI's output is wrong, incomplete, or unsuited to the actual problem, and the domain knowledge to correct it. A student who can operate an AI writing tool but can't tell a well-argued essay from a shallow one hasn't acquired a complementary skill; a student who has strong underlying writing judgment and uses the tool to work faster has. The tool amplifies whatever judgment the person already brings to it — it doesn't substitute for building that judgment in the first place.

Discussion

Run this as a seminar-style discussion rather than a question-and-answer drill. These questions are meant to surface genuine disagreement and are close to the kind of case-study discussion used in real labor-economics or career-planning contexts — it's fine, and often more valuable, if the class doesn't reach consensus.

• "If entry-level compression is real, what should someone your age actually do about it — avoid fields where it's happening, or lean into them anyway for a different reason?"
• "Autor's framework says to evaluate automation task-by-task rather than job-by-job. Pick a career you're considering and try to name three of its component tasks, then rate each one's automation exposure. Does the exercise change your view of the career?"
• "The skill premium is shifting toward 'complementary' skills — those that become more valuable alongside AI. What's a complementary skill you could start building right now, in high school, before you've even chosen a career?"
• "Should schools change what they teach because of this shift? What would you personally want a high school curriculum to include that it currently doesn't?"
• "The policy debate around AI and employment is about balancing innovation with worker protection. If you were designing government policy, what's one concrete thing you'd do to protect workers without slowing down beneficial innovation?"
• "Some 'new AI jobs,' like certain data-labeling work, are lower-wage and less secure than the term 'AI job' might suggest. Does that change how you think about the 'new jobs are appearing' side of this lesson's argument?"

Quiz Walkthrough

This quiz assumes engagement with the economic vocabulary introduced in the activity above — task decomposition, entry-level compression, complementary skills, and the policy debate. If a student can select the correct option without being able to explain the underlying concept in their own words, treat that as a cue to return to the relevant part of Scene 3 rather than moving on.

Wrap-Up & Extension

Close with: "AI automates tasks, not whole jobs. The skill premium is shifting toward creativity, empathy, and adaptive learning. Entry-level positions face real disruption. The best career strategy combines technical AI fluency with the uniquely human capabilities AI still can't replicate."

Extension activity (25–30 minutes, pairs well with Architect mode's portfolio and career path features): Have each student pick a specific career path they're seriously considering and produce a short written task-decomposition analysis: list 4–6 concrete tasks that make up that job today, rate each task's automation exposure (high/medium/low) with a one-sentence justification, and conclude with a short paragraph on what skills they'd want to develop now to stay valuable in that field over the next decade. This is a genuinely useful artifact for a student's own career planning, not just a classroom exercise, and it fits naturally alongside anything else they're building in an Architect-mode portfolio.

A sharper version for a class ready for it: have students research one real, documented case of AI changing a specific profession (a news article, industry report, or documented labor statistic works) and present a short critique — is the source describing task-level automation or making a broader, less-supported claim about the whole job disappearing? This builds the habit of reading claims about "AI and jobs" critically, which is arguably the single most durable skill this lesson can leave them with, well beyond any specific fact from today.

If your school's counseling office runs any kind of career-day or alumni-speaker program, this lesson pairs unusually well with a short follow-up conversation with a working professional — ideally someone in a field a student is considering — asking them directly which parts of their own job have changed because of AI tools in the last few years, and which parts haven't moved at all. A real, current, first-person answer tends to stick with this age group longer than any statistic in the lesson itself.