Loading prices…
STKR NewsSTKR News0 of 3 free this month
AI

AI tools were supposed to make students think less, but a major university trial found a different result

A massive 19-university trial shows that custom AI tools actually improve STEM student performance, debunking the myth that AI is just a tool for lazy shortcuts.

Originally on CryptoSlate →
AB

Adrian Boysel

Contributor

Nov 20, 2026

4 min read

Photo illustration / STKR News

We have been told for two years that AI is the death of critical thinking. If you listen to the louder corners of the academic world, LLMs are basically sophisticated calculators for plagiarism—a way for students to bypass the actual struggle of learning. But a recent trial involving 19 different universities just flipped that script.

The study looked at applied STEM classes. They split students into groups: one group got access to course-specific AI tools immediately, while the other had to wait. The result wasn't a race to the bottom. Instead, the students with the AI tools outperformed those without them. This isn't just about grades; it is about how we actually interface with complex information.

The Shortcut Fallacy

In the builder community, we often talk about 'lowering the barrier to entry.' In crypto, that means making a wallet that doesn't feel like a math exam. In AI, it means removing the friction between a question and an answer. The fear has always been that if you remove the friction, you remove the learning.

What this university trial suggests is that we’ve been looking at friction the wrong way. Most of the 'struggle' in learning STEM isn't the high-level conceptual work; it is the mechanical troubleshooting. It is getting stuck on a syntax error for three hours or failing to understand a single bridge between two complex formulas. When an AI tool acts as a specialized tutor, it clears the mechanical hurdles so the student can spend more time on the actual logic.

Personalized Iteration at Scale

For founders building in the EdTech or AI space, the takeaway here is about specificity. These students weren't just given a blank ChatGPT prompt and told to 'go figure it out.' They were given course-specific tools. This is a massive distinction.

Generic AI is a distraction. Specific AI is a lever. When a tool understands the context of the curriculum, it doesn't just give the answer—it guides the process. We are moving away from the era of 'Ask me anything' toward 'Help me build this specific thing.' That is where the real value is being captured.

  • Context is King: The students who outperformed their peers had tools that were tuned to their specific syllabus.
  • Efficiency vs. Laziness: Using AI to solve a bottleneck isn't cheating; it's optimization.
  • Real-time Feedback: The biggest gap in traditional education is the delay between asking a question and getting a response. AI closes that loop.

What This Means for the Builders

If you are building in AI right now, stop trying to build the 'everything app.' The data from this trial shows that the win is in the vertical. A tool that helps a mechanical engineering student understand fluid dynamics is infinitely more valuable than a general-purpose bot that can write a mediocre poem and also solve a math problem.

We are seeing a shift in the 'proof of work' for human intelligence. In the past, the work was the manual labor of research and rote memorization. Now, the work is the ability to direct the machine, verify the output, and integrate it into a larger system. The students in this trial aren't thinking less; they are thinking at a higher level of abstraction.

The fear that AI makes us stupid assumes that human intelligence is a fixed pool. It’s not. When we automate the mundane, we don’t stop thinking; we just find harder problems to solve.

The Skeptic's Corner

Of course, we have to be careful. A trial across 19 universities is a great data point, but it isn't a universal law. There is a risk of dependency. If a student becomes reliant on a specialized tool to get through a STEM course, what happens when they are in a high-stakes environment without that specific UI?

As builders, our responsibility is to create 'training wheels' that eventually come off. The goal of an AI educational tool shouldn't be to stay with the user forever; it should be to level them up so they eventually outgrow the tool's basic functions. If we only build crutches, we are failing. If we build telescopes, we are winning.

The Long Game

This study is a lighthouse for the next wave of AI development. It proves that the 'doom and gloom' narrative around AI in education is largely based on a misunderstanding of how humans use tools. We don't use hammers so we can stop building houses; we use them so we can build bigger, better houses faster.

The same applies to crypto and the decentralized web. We are building tools to abstract away the complexity of finance and data. If we do it right, people won't 'know less' about money; they will be able to do more with it because they aren't bogged down by the archaic plumbing of the old system.

The Takeaway: Stop worrying about AI making people lazy. Start building specific, context-aware tools that remove mechanical friction. The data shows that when you give people better tools, they don't do less—they achieve more. The builders who win the next decade will be the ones who focus on specialized utility over general noise.


Read the original at CryptoSlate →

The Brief

Stay Updated on Cutting-Edge Tech

A six-minute morning dispatch on the markets and the technology shaping them.

Free. No spam. Unsubscribe anytime.

Write for STKR

Become a Contributor

Earn $STKR for published stories on markets, protocols, and culture.

  • Earn $STKR for every published piece
  • Editorial support from the STKR desk
  • Byline visibility across the network
  • First look at the upcoming creator program
Apply to Write

Keep reading

All stories

Comments

24 reader responses