--- title: "How to Work With AI as a Junior Employee | Openskill" description: "How you work with AI as a junior employee shapes results more than skill scores do. Find your profile." url: "https://openskill.ai/blog/ai-apprentices-delegators-amplifiers/" --- [← Back to blog](https://openskill.ai/blog/) · [AI Careers](https://openskill.ai/category/ai-careers/) # AI apprentices, delegators and amplifiers: which one are you? Sep 30, 2026Updated Sep 30, 20266 min read[Openskill team](https://openskill.ai/about/) [AI upskilling](https://openskill.ai/tag/ai-upskilling/) · [early career](https://openskill.ai/tag/early-career/) · [AI at work](https://openskill.ai/tag/ai-at-work/) · [judgment](https://openskill.ai/tag/judgment/) ### In this article 1. [What the study measured](https://openskill.ai/blog/ai-apprentices-delegators-amplifiers/#what-the-study-measured) 2. [Three ways of working with the same AI](https://openskill.ai/blog/ai-apprentices-delegators-amplifiers/#three-ways-of-working-with-the-same-ai) 3. [Why good fundamentals weren’t enough](https://openskill.ai/blog/ai-apprentices-delegators-amplifiers/#why-good-fundamentals-werent-enough) 4. [How to tell which one you are](https://openskill.ai/blog/ai-apprentices-delegators-amplifiers/#how-to-tell-which-one-you-are) 5. [Moving toward the amplifier habits](https://openskill.ai/blog/ai-apprentices-delegators-amplifiers/#moving-toward-the-amplifier-habits) If you started your career in the last couple of years, you’ve probably been handed an AI tool and a vague instruction to use it. Most likely, the instruction stopped there. A new study of early-career professionals gives a clearer answer than most, and it has less to do with what you know than with how you steer the tool. ## What the study measured KPMG and the McCombs School of Business at the University of Texas at Austin ran a field study of 523 early-career professionals, published in [Harvard Business Review](https://hbr.org/2026/07/research-why-some-junior-employees-work-well-with-ai-and-others-dont) in July 2026. Participants completed assignments with an AI agent that closely mirrored the client work they’d do in their own specialty. The design had one clever step. Before any person touched the task, the researchers had the AI agent complete it alone. That gave them a baseline, the quality you’d get with no human involved at all. Then they compared each person’s result with that baseline. The question became whether a human improved the AI’s work, left it the same, or dragged it down. ## Three ways of working with the same AI The results split people into three groups. AI Amplifiers, about half of the sample, beat the AI baseline. They defined the problem before handing it over, anchored the work in established frameworks from their field, and refined the output across several rounds. They treated the AI like a capable colleague who needed direction and review. AI Delegators, roughly a quarter, produced results about as good as the AI alone. They accepted the first output with little scrutiny and added little of their own. These people scored lowest on foundational skills, yet they weren’t the weakest performers, because the AI’s default output was already competent. AI Apprentices, the remaining quarter, landed below the baseline. This is the group that should make you stop and think. Apprentices matched Amplifiers on every foundational skill measure and scored higher than Delegators. They also pushed back on the AI and critiqued its work. Their critiques just rarely helped, often chasing issues that didn’t matter or steering the AI in the wrong direction. ## Why good fundamentals weren’t enough The Apprentice result is the heart of the study. On paper these were strong junior hires. They knew their material, and they followed the standard advice about AI, which is to question it. They still ended up worse off than if they’d left the AI alone. The lesson is that critique only helps when it’s aimed. Telling an AI “this isn’t quite right, try again” or fixating on the wording of a paragraph while a flawed assumption sits in the first line can pull a decent draft toward a worse one, so scrutiny without a clear frame ends up as noise. Rahsaan Shears, who leads AI transformation at KPMG US, summed it up by saying that “how people applied their knowledge and skill is what made the difference.” That makes the Apprentice group the encouraging part of the study. They already have the subject knowledge, and what they lack is a method for directing the tool, which takes far less time to pick up than a new field. ## How to tell which one you are You won’t get a score, but your own habits give you a decent read. Look at the last few tasks where you used AI and ask yourself some honest questions. Did you define the problem before you prompted? If your opening message was the task pasted in with “help me with this,” that’s the Delegator starting point. Framing the problem the way Amplifiers did means spelling out the goal, the audience, the constraints and what a good answer would include. How much of the final version came from you? If you’d be hard pressed to point to anything you changed beyond small edits, that’s the Delegator pattern. It feels efficient, and your manager may not notice right away, but it caps you at what anyone with the same tool could produce. When you pushed back, did the output improve? This is the Apprentice test. Compare the first draft with the last. If you went through several rounds of feedback and the final draft isn’t clearly better, or you can’t explain the improvement, your critiques weren’t landing. ## Moving toward the amplifier habits The study describes what Amplifiers did in fairly concrete terms, and each part turns into a practice you can start this week. Frame the problem first. Before you open the tool, write two or three sentences on what you’re trying to decide or produce, who it’s for, and what would make it wrong. Paste that in with the task. A clear frame usually does more for quality than prompt tricks, and it forces you to think before the AI does. Bring a framework from your field. The Amplifiers anchored their work in domain frameworks, the established methods professionals in their area use to structure a problem. In audit that might be a risk assessment approach, in marketing a segmentation model, in engineering a set of design principles. Tell the AI which one to apply, then check that it followed it. This is where your education pays off, because a model will happily produce generic work unless someone who knows the field insists on rigor. Make your critiques targeted. Before you give feedback, decide which single issue matters most to the result. Name it specifically, explain its impact, and describe what a fix would look like. “The revenue forecast assumes last year’s churn rate, but we changed pricing in March, so rerun it with the new rate” gives the AI something to act on. “Make this better” sends it wandering, and feedback that misses the point is what held Apprentices below the baseline in the study. Iterate with a purpose, and know when to stop. Amplifiers refined results across multiple rounds, so give each round one clear aim. If two rounds of feedback haven’t moved the work forward, rewrite the weak section yourself or step back and reframe. If you want a related view on how employers now think about this, our piece on [supervising AI output](https://openskill.ai/blog/employers-want-you-to-supervise-ai/) covers what hiring teams look for. The story of how you set up a task, which framework you applied and what you changed in the AI’s draft is what separates an Amplifier from someone who simply used the tool. [Openskill](https://openskill.ai/) runs the most realistic AI voice interviews, with follow-up questions based on your answers, so you can practice explaining how you framed and critiqued AI work before a hiring manager asks. Paste in the job description for the role you want, and the session is built around that role. If you are also job hunting, [openskill](https://openskill.ai/) covers the rest of the loop. Search current openings in [Jobs](https://app.openskill.ai/jobs), open a role and click “Prepare for this interview” to practice for that exact job. ## Questions, *answered.* ### What are AI Amplifiers, Delegators and Apprentices? They're three profiles from a 2026 KPMG and University of Texas at Austin study of 523 early-career professionals. Amplifiers, about half the group, beat what the AI produced alone. Delegators matched the AI baseline by accepting its output, and Apprentices critiqued the AI but steered it the wrong way and landed below the baseline. ### How should a junior employee work with AI? Frame the problem before prompting, anchor the work in a framework from your field, and refine the output through specific, relevant critique over several rounds. In the study, those habits separated the people who beat the AI from those who matched or trailed it. ### Do foundational skills still matter if AI does the work? Yes, but they aren't enough on their own. In the study, Apprentices scored as well as Amplifiers on every foundational skill measure and still performed below the AI working alone, because their feedback pushed the AI in unhelpful directions. ### What is the best way to practice interview answers about working with AI? Pick one task where you framed the problem, applied a method from your field and corrected the AI's draft, then explain it to someone who asks follow-ups. Openskill runs the most realistic AI voice interviews and can start from a job description, so the follow-up questions stay close to the role you're applying for. Compare tools ## Choosing where to practice? See how Openskill compares with other interview practice tools, and how its job search and AI match score fit alongside realistic voice practice. - [Best AI mock interview apps (2026)](https://openskill.ai/best-ai-mock-interview-apps/) - [Best mock interview apps for job-specific practice](https://openskill.ai/compare/best-mock-interview-apps-for-job-specific-practice/) - [Best realistic voice AI mock interview apps](https://openskill.ai/compare/best-realistic-voice-ai-mock-interview-apps/) - [Best all-in-one AI interview prep platforms](https://openskill.ai/compare/best-all-in-one-ai-interview-prep-platforms/) - [All interview tool comparisons](https://openskill.ai/compare/) Share[X](https://x.com/intent/post?text=AI%20apprentices%2C%20delegators%20and%20amplifiers%3A%20which%20one%20are%20you%3F&url=https%3A%2F%2Fopenskill.ai%2Fblog%2Fai-apprentices-delegators-amplifiers%2F) · [LinkedIn](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fopenskill.ai%2Fblog%2Fai-apprentices-delegators-amplifiers%2F) Keep reading ## Related *reading.* AI Careers ### [The AI habit most people skip is questioning the output](https://openskill.ai/blog/question-ai-output-fluency-habit/) Anthropic's AI Fluency Index found that most people refine AI answers, but they question reasoning and spot missing context less often once the output looks finished. A short checklist makes that second habit automatic, especially for documents and code. Oct 10, 2026 7 min AI Careers ### [How to build judgment when AI does the junior work](https://openskill.ai/blog/build-judgment-when-ai-does-junior-work/) The routine tasks that used to teach new hires how to make good calls are being handed to AI. A practical guide for early-career people on getting experience with decisions under uncertainty anyway, from owning small calls to studying how seniors made theirs. Oct 8, 2026 6 min AI Careers ### [A 30-day AI upskilling plan for the job you already have](https://openskill.ai/blog/30-day-ai-upskilling-plan/) The fastest way to get good with AI is to use it on the work already on your desk. This week-by-week plan has you map your tasks, redesign two of them, practice daily on real work and then measure and share what changed. Oct 12, 2026 6 min