May 13, 2026
The AI Generation and the Illusion of Understanding
Over the past year, I have received a growing number of emails from high school students expressing interest in regenerative medicine and hearing research. Many of these messages are thoughtful, polished, and enthusiastic. Some students clearly used AI tools to learn enough background to ask intelligent questions.
This is both impressive and concerning.
Artificial intelligence can now generate remarkably good summaries of complex biology. A motivated teenager can learn the vocabulary of stem cells, signaling pathways, and regeneration in a matter of hours. They can discuss EGFR signaling, hair cell regeneration, or transcription factors with surprising fluency.
But fluency is not the same as understanding.
When I learned biochemistry decades ago, the process was slow, repetitive, and difficult. We memorized pathways, drew structures by hand, struggled through problem sets, and were tested without books or computers. Much of the factual detail faded over time, but the intellectual framework remained. That painful process built intuition: an internal sense for how biological systems behave, why experiments fail, and why seemingly simple ideas become difficult in real organisms.
That kind of understanding cannot be downloaded instantly.
Biology is not a collection of keywords. It is a discipline built on layers of chemistry, physics, statistics, uncertainty, and years of exposure to experimental reality. AI can help students access information quickly, but it cannot replace the mental restructuring that occurs when someone wrestles deeply with a problem.
This creates a new challenge for educators and scientists. Young students can now appear far more advanced than they truly are. Some are genuinely curious and will grow into excellent scientists. Others are mainly collecting credentials, affiliations, or lines for a résumé. The difficult part is that AI makes these groups harder to distinguish at first glance.
So what should young students do?
Use AI — but use it as a tutor, not as a substitute for thinking.
Read real papers, not only summaries. Learn enough chemistry and mathematics to understand mechanisms instead of narratives. Spend time being confused. Try to understand one figure deeply rather than ten papers superficially. Learn how experiments are controlled and why biological systems resist simple explanations.
Most importantly, develop patience.
Real scientific understanding grows slowly. The goal is not to sound intelligent. The goal is to build a mind capable of recognizing what is true, what is uncertain, and what remains unknown.
That process still takes years. AI has not changed that.
A personal note: AI also makes me sound more intelligent and eloquent than I ever did on my own. I was a C–D student in German, with occasional highlights, a clear F in Latin, and barely passed high school English. So how can I write polished blog posts like this? Guess.
AI helps. A lot.
But the core ideas still have to come from me. They need to marinate. They need to connect to lived experience, scientific training, frustration, curiosity, and memory. Then they can be drafted, revised, and polished. AI is excellent at polishing language, and sometimes it even sharpens the thinking. I value it.
Scientific writing remains a trained skill. I am grateful to my teachers and mentors - Jim Hudspeth, of course, stands out - and to many others who gave feedback over the years. That long process mattered.
So yes, AI is a great tool. But it does not replace the slow building of mental structures that allow concepts to connect to one another. It can improve the sentence. It cannot substitute for the years required to know what the sentence should say.