Technology leadership, classroom exposure and building the next generation of practitioners

Thoughts of a Tech CEO: What We Owe the Next Generation.

A personal leadership perspective from Jeffery Williams on the responsibility technology professionals share: help students encounter real careers, build durable foundations, question AI output and imagine a future they can help shape. An hour at a school or library can be a practical place to begin.

JWilliams & Associates · November 3, 2026 · 11 min read

An invitation for technology executives, engineers, and business leaders to invest time in the people who will inherit the industries we're transforming.

A personal perspective by Jeffery Williams
CEO, JWilliams & Associates

I have spent decades developing enterprise systems. Today, I work directly with agentic AI: software that can use tools, follow a sequence of steps and take actions inside a workflow. I can see how much that changes the questions we need to ask about work, judgment and responsibility.

It also brings a responsibility beyond the next application or release. The people who will inherit these systems are learning about the world right now. How much of our experience are we making available to them?

The thought that keeps returning

Who will understand the systems we are building well enough to question them, improve them and decide when they should stop?

Operating a tool is one capability. Understanding why it behaves as it does, recognizing an unsupported answer and accepting responsibility for a consequential decision require more. Those abilities take practice, knowledge and judgment.

They develop over years. Curiosity gives that development a place to start.

We have an opportunity to contribute while those habits are forming. In classrooms, libraries and career programs, students are deciding what interests them and what they believe might be possible. People who build technology can help make some of those possibilities visible.

The opportunity gap is already visible

The need for that connection is more than an executive’s intuition.

The OECD’s 2025 report on teenage career preparation found that 39% of students across OECD countries were unclear about their career expectations. That measure comes from 15-year-olds surveyed in PISA 2022. The report also describes unequal access to career preparation and the influence of social background on educational expectations.

That finding makes me ask whose opportunities depend on already knowing somebody in the industry. A student whose family has no connection to engineering should still have a chance to discover what engineers do.

The OECD’s guide to career talks draws on longitudinal research linking some forms of career exposure with better later employment outcomes. Those associations do not mean one visit determines a student’s future. They give us a reason to take thoughtful, sustained opportunities for contact seriously.

Pillar one: exposure makes the unseen tangible

It is harder to imagine a career when you have never met anyone who does the work or seen what solving its problems feels like.

An industry becomes more tangible when a person explains the questions behind the job: why a warehouse process gets delayed, how a medical device is tested, or what an engineer does when the information is incomplete. Students can ask questions a polished career description never invites.

Show the reasoning, the collaboration and the revisions. Show that uncertainty is part of the work and that competent people check their assumptions. A student should leave with a clearer picture of a profession and permission to be curious about it.

Pillar two: foundations outlast the tools

Our contribution also needs to look beyond the software we happen to use this year.

Learning today’s tools can be useful. A preparation built only around their current interfaces is fragile. Models, products and workflows change, and a student’s education needs to support a much longer life than a release cycle.

Reading carefully. Understanding mathematics, measurement and probability. Communicating a thought clearly. Analyzing evidence. Working through an unfamiliar problem. Considering who could be affected by a decision. These foundations give students more ways to learn when the tools change.

They require teaching and practice within real subjects. The Education Endowment Foundation’s guidance recommends explicitly modeling how to plan, monitor and evaluate learning, with support that helps students do more independently. Its emphasis on subject content matters: asking students to “think critically” is an invitation; demonstrating how to test a mathematical claim or revise an argument makes that invitation usable.

Teach young people to understand problems deeply enough that they can adapt when the software changes. That is the preparation I want us to help strengthen.

Pillar three: AI literacy is judgment, not tricks

Useful AI literacy includes knowing when a tool helps, how to examine its answer and when to set it aside.

The version worth teaching is harder and more useful. How does a system produce an answer? What is it doing when it retrieves versus when it generates? Where does it fail, and how would you know? What data did it see, who approved that, and who is accountable when it is wrong?

UNESCO’s student AI framework places a human-centered mindset and ethics alongside understanding AI techniques, applications and system design. It treats students as people who can question and help shape these systems. That is a broader ambition than learning a set of prompting techniques.

I want a young person to be comfortable asking: Where did this information come from? How could I check it? What might be missing? Who should make this decision? Those questions have value well beyond an AI interface.

What the classroom research suggests, and what it does not

A classroom study offers a useful example of why educational design matters.

In a 2025 PNAS paper, Bastani and colleagues studied nearly a thousand mathematics students at one Turkish high school using GPT-4-based tools. Access to a general-purpose assistant improved assisted practice performance, while later unaided performance could suffer. A more constrained tutor, designed around hints and teacher-provided material, largely mitigated that harm.

The setting was one school, one subject and an earlier model generation. It does not settle what every AI tool will do for every student. It does show why finishing an assignment and developing the ability to do the work deserve separate attention.

I would want the same distinction made in any learning exercise: what did the tool accomplish, and what can the student now explain or do independently?

The OECD’s Digital Education Outlook 2026 similarly cautions that better task performance does not necessarily produce learning gains. It emphasizes purposeful educational use and the development of independent thinking. We should bring that distinction into the classroom conversation, along with the possibilities.

Pillar four: every industry, not just software

A student’s interest in technology can begin inside another profession. We should make room for that.

A nurse examining a system’s recommendation, a farmer interpreting sensor readings, a machinist investigating an exception and an electrician evaluating a control all bring domain knowledge to technical questions. These are examples of how judgment and technical literacy can work together, rather than predictions about any particular job.

Bring people from healthcare, manufacturing, finance, education, agriculture and skilled trades into the conversation. Let students see several routes: academic study, technical education, apprenticeships and continued learning at work. Their interests should have room to develop.

Knowing a field deeply enough to ask a better question is a strength. We should help young people recognize it.

Pillar five: make participation accessible

The first contribution can be small and specific.

You do not need to endow a scholarship. You do not need to start a foundation. You do not need a program with your company's name on it. You need an hour, arranged with a school or library that already exists, and the discipline to actually show up.

One visit. One conversation. One honest account of what your work is really like, including the parts that are tedious.

A classroom talk can be a useful one-time commitment. Mentoring calls for a different promise: continuity, agreed boundaries and participation through the host’s established arrangements. Choose what you can honestly sustain, then do what you agreed to do.

Teachers are partners, not the problem

We should arrive with respect for the professionals who are already there.

Teachers know their students, curriculum and constraints. Our contribution is practical subject knowledge and a view of the work. Their expertise should shape how we share it.

Arrange activities in advance, follow the host’s safeguarding and participation rules, and adapt to the students’ ages and needs. UNESCO’s education guidance emphasizes age-appropriate design and privacy. A demonstration can use fictional information and an approved adult-operated tool, without requiring students to create accounts or disclose personal information.

It also means listening first. Ask what would help. Frequently the answer is not what we assumed.

Something you can do in a single class period

Here is one possible 30-to-60-minute activity to adapt with a teacher or librarian.

Start with a small problem, such as planning a community garden within a fictional budget. Ask what information is needed, which units matter and how the group would check the total. Let students reason through it and explain an initial answer before introducing AI.

Show a prepared response from an approved tool, or a worked example on paper. Compare it with the supplied prices and constraints. Is every claim supported? Were the quantities and units handled correctly? If the answer is correct, explain how it was checked. If it is wrong, identify the evidence that reveals the error.

Then ask students to revise their answer where needed and explain the reasoning in their own words. Invite them to say what evidence would make them change their minds. The conversation about checking can matter as much as the final total.

A printed example with a deliberate error or a physical model can support the same exercise when a live tool is unsuitable. The host can adjust the difficulty and connect it to the subject being taught. Curiosity and participation should not depend on owning a device or paying for an account.

The ten- and twenty-year view

I am skeptical of confident labor forecasts, including the ones from my own industry. Nobody knows what the 2036 or 2046 economy looks like in detail, and I will not pretend to.

What I can reason about is what kind of person thrives under that uncertainty. Someone who can take a domain and combine it with technical reasoning. Someone who keeps learning because they are used to it. Someone with the ethical instinct to ask who is affected and the communication skill to explain it to a room that disagrees.

No single lesson makes a child ready for an entire career. Reading, mathematics, communication, collaboration and ethical judgment develop through repeated learning and practice. Encounters with working professionals can give that effort a context students care about.

Our role is to add useful experiences and open possibilities while allowing young people to choose paths we may never have imagined.

What I am asking, specifically

This month, contact one school, public library or career program in your community. Ask what help would be useful. Consider organizations with fewer existing connections to your industry.

Then agree to a specific hour. Put it in the calendar. Listen more than you present. And follow through.

Start with a commitment the host actually wants and you can keep. If a colleague is interested, encourage them to arrange their participation with the organizer too. A contribution does not need a branded initiative to be worth making.

I see this as part of leadership: making our experience available to the people whose questions and choices will shape what comes next. We can contribute without deciding their ambitions for them.

Why this is worth our time

Students deserve opportunity for more than their future contribution to a workforce. They deserve the knowledge and confidence to understand systems that affect them, question decisions and participate in shaping their communities.

That is a reason to give our time even when the benefit will never appear in our own business results.

The technology we build today may be obsolete in twenty years. The curiosity we inspire in a student today could influence what gets built fifty years from now.

We don't need our names attached to that contribution. We simply need to make it.

Sources & references

OECD Digital Education Outlook 2026
OECD Digital Education Outlook published in January 2026, reporting emerging research indicating that generative AI can improve performance on a task without ensuring that learning occurs, and emphasizing that educational design and teaching intent shape the outcome, with a focus on developing independent thinking and selective, purposeful use rather than replacing effort or human relationships. Cited as a directional finding from emerging research; it is not presented as a settled universal conclusion.
AI competency framework for students — UNESCO
UNESCO's 2024 AI competency framework for students, organized around twelve competencies across four dimensions — a human-centred mindset, ethics of AI, AI techniques and applications, and AI system design — with a progression from understanding through application to creation, and an emphasis on critical judgment, responsible citizenship and foundations for lifelong learning rather than tool operation. Cited for that framing only; no claim is made that the framework is a legal requirement or that it has been adopted by any school system.
Metacognition and self-regulated learning — Education Endowment Foundation
Education Endowment Foundation guidance (second edition, 2025) on metacognition and self-regulated learning, recommending that teachers model thinking explicitly, scaffold planning, monitoring and evaluation of learning, and embed these strategies within subject content rather than treating them as generic skills taught in the abstract. Cited for that practical point about how the capability is built; no specific attainment figure or transfer claim across all subjects is asserted.
Guidance for generative AI in education and research — UNESCO
UNESCO guidance advocates a human-centered, age-appropriate approach and protection of privacy in educational AI use. Used for adapting demonstrations with educators, not to assert universal legal requirements.
Generative AI can harm learning (Bastani et al., PNAS)
Peer-reviewed field experiment by Bastani and colleagues, published in PNAS in June 2025, conducted with nearly 1,000 students at one Turkish high school during the 2023–24 academic year in mathematics, comparing GPT-4-based conditions including a general-purpose assistant and a more constrained tutor design against control. The study reports that a general-purpose assistant improved performance on practice exercises while in the studied conditions reducing later performance when students worked without the tool, and that the more constrained, hint-based tutor working from teacher materials largely mitigated that harm. Cited narrowly: one school, one subject, one model generation; no general claim about all AI, all students or all subjects is drawn from it.
The State of Global Teenage Career Preparation — OECD
OECD report published in 2025 drawing on PISA data from 2000 to 2022 across more than 80 countries, finding that a significant portion of students are uncertain about their career plans, that this uncertainty has grown since 2018, that students participating more in career development activities tend to have clearer plans and better employment outcomes, that educational ambitions are more strongly determined by socio-economic background than by academic ability, and that employer engagement and effective career guidance matter while too few students participate in the activities most strongly linked to better outcomes. Cited for those findings; no causal guarantee from a single engagement is asserted.
Custom AI development — JWilliams & Associates
Supports only the separately labeled publisher advertisement for custom AI applications, integration, permissions, evaluation and operational planning. Not a claim of an educational product, school program or child-facing service.
Career talks with guest speakers — OECD
2023 OECD guide reviews career talks and longitudinal associations with later employment outcomes. Used as evidence for organized career exposure, not a guarantee that a single visit causes a particular result.