Close your eyes. Imagine: What do you think education will look like in five years? How about in ten years?

For most of us, education, while having undergone incremental changes, has mostly stayed the same. We picture students sitting at desks with a teacher at the front of a classroom in a brick and mortar school. Students are grouped into grades largely by age. They may take yellow school buses to and from their homes. Minus the laptops, smart phones, and a few other tech gadgets, the environment would not be totally alien to someone from many decades past. We have an opportunity to change all of that and afford students an individualized learning experience that meets them where they are, regardless of age or grade level. And we can do it for little to no cost (maybe even with some cost savings).

Sam Altman, CEO of OpenAI, just observed that while the growth of computer technology has improved 2x every 18-24 months for more than fifty years, AI performance has demonstrated a staggering 150x improvement over the last 18^th months. Most strikingly, is that this trend appears to be accelerating even faster. Recently, the DeepSeek AI demonstrated recursive self-improvement. It was quite literally prompted to find a way to improve its performance and it managed an astonishing 2x speed increase almost immediately. We have a tool in our hand which knows how to improve itself. For many observers this is a milestone for even more rapid development of technology in general, and AI in particular.

The 21^st century has already transformed the workplace and the tools we use every day. The information age has radically changed what skills are most valuable for many industries. Memorization is no longer the core skill and going forward into the AI age we will see another transformation. The education system was slow to adapt to the first transformation, we should be faster to adopt the second.

Estimates wildly vary as to the number of jobs this will displace or augment, but many think that AI will create more jobs than it destroys. Personally, as a technology futurist, I think these are conservative estimates. It is more likely that all jobs and aspects of life will be profoundly impacted–including schools and education. As Vermont taxpayers are struggling with massive costs due to excessive staffing, which place us as the 2nd highest per-student spending in the Country, we could use some really disruptive thinking about how to control costs, while simultaneously delivering personalized world-class education. AI, in and out of the classroom, might be part of the solution.

I recently had the pleasure of speaking with Professor Stephen Kosslyn, a psychologist and neuroscientist, who spent 30 years researching and teaching at Harvard University. Kosslyn has penned multiple books on the subject of online and AI-assisted learning. His approach advocates for having AI function as a cognitive amplifier. What this means is that the AI is not meant to think for us, but to draw out our capacity for deep learning and critical thinking – a personal learning coach as it were.

He advocates for a process of learning by using, capitalizing what is sometimes referred to as the generation effect. If I can employ what I’ve learned in real-world applications, if I can express it, not as a matter of rote memorization for a test, but I can use what I’ve learned again and again, then I’ve achieved mastery and retention.

Part of the problem with the traditional classroom is that it rests on the presumption that students of the same age should be able to cover the same material at more or less the same pace. Each student spends the same amount of time on a given learning lesson, but achieves variable results. This generates endless frustration as many areas of learning are cumulative, so if students missed something in math last month or last year, the gap in understanding may come back to haunt them as they attempt to move forward.

Personalized learning aims to find that sweet spot with each student, the middle ground which is a stretch and challenging, but not so overwhelming as to plunge the student into an anxiety-induced panic, nor too slow and easy as to be boring. Teachers under the current classroom configuration are forced to teach to the average student, leaving some students either overwhelmed (class is moving too fast) or disengaged (the class is moving too slow). But what if we freed up teachers to become guides who can walk around the classroom and check in on students who are all in different places, students being challenged with a pace that is tailored to where they are in their individual process?

AI assisted learning is now capable of becoming a lifelong copilot for every person. Imagine everything a person has ever attempted to learn being tracked with all of the difficulties and gaps in knowledge noted. By thinking of learning as modular, the AI could offer a timely refresher on fractions, or a much-needed review of the proper use of a semicolon, and then immediately provide that lesson using multiple teaching styles aligned with the student’s particular interests. It would essentially create individualized learning plans (IEPs) for each student. Gone would be the days of students hiding in the back row of the classroom or spacing out and daydreaming while other students or the teacher speaks.

Active learning means each student is prompted to engage the material throughout the lesson via their computer in or out of the classroom. There would no longer be any possibility of passivity. Performance would spike. In addition to this, Kosslyn has demonstrated in his work how critical thinking, creative problem solving, emotional intelligences, collaboration and communication may be cultivated in unprecedented ways.

Beyond achieving a more desirable student/teacher ratio, the price performance is the best part of this. Kosslyn informed me that aside from the curriculum creation, which can run from $10k to over $100k per course, (depending on length and if his team has to start from scratch or build off of pre-existing materials), the actual price per student achieves a near zero marginal cost. This means that it does not matter whether you provide the AI assisted learning to 1,000 students or 100,000. Once you have the material and computers, the addition of each new student may add less than $100 to a year-long course on any given subject.

All of this is not to suggest that AI will replace teachers. Far from it. Rather AI assisted learning can greatly extend what a single teacher can accomplish, focusing the teacher’s time on getting to know the students better, and spending less effort trying to control the classroom. Teaching has the potential to shed much of the repetitive drudgery and become refocused on the fine art and creativity of student mentoring, deepening personal connections and understanding, while fine-tuning the learning process.

What is most exciting is that this novel reality in education is not a distant dream which is five or ten years away. It is here now. At Campaign for Vermont, we believe it’s time to run a pilot program at a small group of schools to prove the application of individualized learning, via AI coaches, that is urgently needed to prepare us for the jobs of the future.

We can achieve new educational heights without draining our pocketbooks.

Asher is a Technology Futurist who speaks and consults globally for a wide range of companies. In addition to being a policy researcher and advisor to CFV, he is also a ninth generation Vermonter