I have had AI agents draft implementation plans that looked perfect. When I reviewed them, I knew they were not feasible.
Catching that is one part of what an AI Amplifier does. An AI Amplifier is someone whose work with AI turns out better than what the AI produces alone. They frame the problem, add the context AI lacks, push back where it is wrong, and build on what it does well.
I have also seen AI misquote what a customer said and misread the statistics in an analysis. Once, it built a startup's strategy on numbers from a giant established company, where the realities are completely different. The work looked finished in every case. It took an expert review, by someone who knew the business, to see the problem and turn the draft into something we could use.
The idea reflects how RightSeat has worked from the start: people and AI should strengthen each other's work.
A recent KPMG and University of Texas at Austin study1 puts numbers behind that idea, and it found something surprising. Some skilled professionals who worked with AI produced worse results than the AI produced on its own.
The July 2026 study followed 523 early-career KPMG professionals in the US as they completed tasks like their client work. Researchers compared each person's results with what the same AI produced with no one guiding it.
Just over half did better than the AI alone, and the researchers called them AI Amplifiers. About a quarter did about the same as the AI. The last quarter did worse.
That last group was just as skilled as the Amplifiers on paper. They engaged, questioned the AI, and tried to improve its work, but their effort often went to minor issues or steered the work in the wrong direction. KPMG called this group the biggest pool of untapped potential and described the gap as coachable.
The study covers one firm's early-career staff, so it is an early signal. In our view, the skill it points to belongs in employee development at every career stage: knowing which parts of AI-generated work to question and what to add.
For business leaders, this is where the return on AI is decided. Much of the time AI saves can disappear into fixing its output. In a 2026 survey commissioned by Workday2, nearly 40 percent of AI time savings went to rework, and only 14 percent of 3,200 employees consistently came out ahead. Those who did were far more likely to have had skills training.
In our view, that is why AI Amplifiers matter. They direct the work well from the start, so less of the time AI saves is lost to rework.
Most organizations already have a starting point. The people who volunteer to test new tools, answer colleagues' questions, and share what works are often your internal AI champions. Developing them into Amplifiers builds judgment they can share with colleagues.
The KPMG study looked at early-career staff. In our view, there is also a large opportunity with people who already know the work. A veteran proposal manager knows which commitments the delivery team can keep. A senior accountant knows which discrepancies deserve a closer look.
AI is not an oracle. It may not know your delivery limits, your exceptions, or your history. Experienced employees bring that context, and it pays off when they use it to direct and review AI's work.
When we work with customers, we help people build the habits that make them Amplifiers in their own roles. With a proposal team, for example, we teach people to give AI the buyer's priorities before it drafts, then check the draft against what the buyer asked for. They question claims with nothing behind them and confirm the company can deliver what the proposal promises. Good feedback points to a specific gap, such as a requirement left unanswered or a delivery date no one has confirmed.
Today's business students will lead organizations where AI produces much of the first draft: the analysis, the forecast, the proposal. Most competitors will have access to similar tools. In my work advising leaders on AI strategy, the advantage goes to organizations whose people can judge AI's work, know when to push back, and decide what is ready to reach a customer.
Those leaders will set the standard for how their teams use AI, so the habits they form now matter. Students have less experience to draw on, but college gives them a lower-stakes place to start. A wrong answer costs a practice question, not a client relationship, and they can grow into Amplifiers as their expertise develops. Experienced employees start from the other direction, with deep knowledge of the work and a need to practice applying it to AI. The skill is the same at every career stage; only the starting point differs.
Employers are already asking for this. In NACE's 2026 employer survey3, employers reported that more than one-third of entry-level jobs require AI skills, nearly triple the share six months earlier.
That idea shapes our fall 2026 workshop series for students at Clemson University's Powers College of Business, led by RightSeat co-founder Chase Hornbecker, a Clemson senior who teaches his peers. In one session, Chase showed students how to use ChatGPT, loaded with their own course materials, as a study partner that keeps them doing the thinking. Students answer each practice question before ChatGPT explains it, work out why each answer choice is right or wrong, and use the questions they miss to decide what to study next.
Research supports keeping students in charge of the thinking. At Harvard, an AI tutor that walked about 200 physics students through each problem step by step produced more than twice the median learning gain of an active-learning class4. In Turkey, nearly 1,000 high school math students practiced with a standard AI tutor. They did better in practice, then scored 17 percent lower on a later exam without AI than students who practiced with no AI5. A tutor that gave hints and held back full answers avoided that drop. Together, the studies suggest that how AI is designed to support a student's thinking matters.
Our aim at Clemson is for students to practice owning their reasoning while using AI, and to carry that habit into their first job. Employers can reinforce it from day one by asking new hires how they reached a result and what they checked.
Usage reports and training attendance tell you who is using AI. They do not tell you whether the work got better.
Employees may feel behind on AI or worry about what it means for their jobs. Few have hours to spare for training, and in our experience, an hour-long session rarely changes how people work. Short, hands-on practice on approved tasks from their own role builds confidence along with the habits that matter: what to question, what context to add, and when to push back.
If you are paying for AI and want a better return on that investment for your teams and business, start by developing AI Amplifiers in one role. We will help you choose the task where judgment matters most, build practice into the work your team already does, and show you whether the results improved. For colleges, we bring the same hands-on approach to the classroom. Talk with us about developing AI Amplifiers on your team or in your classroom.