Recently, the Fresno Unified School District announced a major AI initiative. Trained teachers are now on hand at sites across the district to assist in implementing the program.
But could it be that the district has made a questionable choice in embracing AI so wholeheartedly?
Jared Cooney Horvath (Ph.D., M.Ed.) would certainly think so.
In The Digital Delusion: How Classroom Technology Harms Our Kids’ Learning—And How To Help Them Thrive Again, the neuroscientist reviews academic research that’s been done so far regarding EdTech (educational technology). To improve K-12 education, he contends, schools need to move along a significantly different path.
His prologue opens with a disturbing statement: “Our children are less cognitively capable than we were at their age.”
Well-developed cognitive skills correlate with “better health, longer life, more stable relationships, higher income and greater life satisfaction.” Meaning that a lot is at stake when such abilities decline.
During the 20th century, cognitive skills trended upward. But that trajectory shifted dramatically when the 21st century got underway. That’s when IQ statistics started to fall in Western nations.
That downward shift didn’t happen everywhere, though. In nations where traditional educational practices continued for the most part, IQ scores are still improving.
Horvath sees a direct connection between the widespread use of digital tools in educational settings and this cognitive decline.
To be sure, the rise of digital education has been a boon for business.
Today, EdTech is a $400 billion sector that’s deeply embedded in many parts of mainstream school culture.
What accounts for EdTech’s marketing success? In part, it’s because its proponents have repeatedly contended that schools as traditionally configured have failed abysmally and that they desperately need a major overhaul.
However, as Horvath points out, available evidence doesn’t support this gloomy assessment. Graduation rates have been going up, and educational inequities among racial and ethnic groups have been falling. In addition, parents and learners have consistently signaled steady levels of approval of their schools.
Despite the promises that EdTech firms made and the high hopes that they inspired, improvements in student achievement just haven’t materialized.
Horvath draws on neurobiology to explain why digital learning hasn’t benefited schools as advertised—and why EdTech products will unavoidably fall short of their promised efficacy in the future.
Educational technology, it turns out, negatively impacts three pillars of learning success: attention, empathy and transfer.
First, consider the importance of paying attention.
When using digital devices on their own, eight- to 18-year-olds have become accustomed to switching rapidly among several online applications. This habit of multitasking weakens their ability to focus—and hence their capacity to absorb new knowledge and skills.
When young learners are asked to work on laptops or other digital tools in classroom settings, they instinctively adopt such a multitasking mindset.
The result, as research indicates, is that on average a student seated in front of a screen will succumb to “unrelated digital distractions” in under six minutes. Overall, students using laptops in class will be off task between 24 and 38 minutes of each hour.
Though EdTech firms insist that they want to enhance education, they are also focused on maximizing the time spent on screens. The two objectives are diametrically opposed to each other, leading to a substantial opportunity cost for schools. Actual learning, Horvath points out, requires “stillness, stability and sustained thought.”
The second pillar of effective learning is empathy.
Research has established that a strong, positive relationship between an instructor and learners is among the most consistent elements of learning success.
Such a relationship bolsters both understanding and motivation.
When two people empathize reciprocally, aspects of their physiology—their brain function, heart rates and breathing—start to match each other.
Which points to a second shortcoming of digital learning tools: they don’t have a biological presence that could make such physiological synchrony possible. Establishing such an empathetic relationship with a computer isn’t just hard; it’s out of the question.
The active presence of instructors is crucial in this regard. “That’s why teachers still matter—not just for what they know, but for who they are. Because in the end, learning isn’t just cognitive; it’s deeply human.”
The third pillar of effective learning is transfer: the capacity to draw on what you’ve learned when you move into new surroundings.
If you acquire information and skills in just one way, your memories will become tied to that setting, and they’ll be less robust. By contrast, if you learn in multiple environments, you’ll usually be able to access what you learned anywhere.
For example, people who study in various ways—at home, in a library, using books as well as screens—will be better able to access knowledge later on no matter where they are.
This points to a third shortcoming of EdTech: Digital tools severely limit the way that learning takes place.
Those who only acquire knowledge via screens will find it harder to access that knowledge in other, novel situations.
Horvath summarizes the incorrigible weaknesses of educational technology this way: “EdTech isn’t failing because of outdated software or poor teacher training. It’s failing because it’s fundamentally incompatible with how human beings actually learn.”
So what about AI?
Rather than viewing AI as an exciting game-changer, Horvath places it within the stream of EdTech products and promises.
And just like many of their EdTech predecessors, AI tools promote cognitive offloading—meaning that users can delegate mental tasks to them. Spell checkers involve cognitive offloading, just as calculators do.
ChatGPT and its fellows, however, turbocharge this behavior.
People who habitually offload tasks will, of course, come to rely on the offloading tool. Eventually, they might well lose the skill to complete such work on their own—or never master that skill to begin with. If learners come to rely on AI help, abilities such as close reading, writing and critical thinking could be hampered or never be developed as they might well be otherwise.
AI platforms appear to offer a shortcut for educational success, but using them extensively in classrooms runs the risk of short-circuiting key components of the learning process.
And so introducing AI into regular classroom practice could well exacerbate the decline in cognitive abilities that EdTech has already helped to bring about.
In the third section of his book, Horvath presents a wide range of strategies for parents, teachers and other stakeholders to push back against EdTech. Among other things, Horvath argues that families should be able to opt out of digital instruction.
Fresno Unified is in the process of establishing an advisory committee to handle issues about its AI implementation. According to the district’s Constituent Services office, this committee will determine whether parents will have such an opt-out choice regarding AI.
As of mid-March, they hadn’t made a decision about that yet.
