As the 2025-2026 school year wraps up, here’s a final pop quiz: How did your understanding of artificial intelligence change over the past nine months?

This year saw AI move from a curiosity to the center of conversation at schools across the country. Many developed or refined policies around AI usage by students and instruction by faculty. And plenty of professional development sessions discussed its potential.

AI understanding, though, remains largely elusive. As we noted in February, many communicators are stuck on baseline questions about the technology, such as which chatbot is “best,” rather than finding ways to use it.

The summer ahead offers us a chance to explore AI and catch up on the latest offerings.

Here are some prompts toward incorporating AI, as well as pitfalls to recognize.

What exactly do we mean by ‘AI’?

In her new book, “I Am Not a Robot,” former Wall Street Journal reporter Joanna Stern points out that Corporate America “is tossing around ‘AI’ as if it were one single, magical brain that can do everything. But AI is just an umbrella term. Underneath it are dozens of different tools, systems, and species.”

Schools are prone to this catchall habit, too. So here’s what Refill means by AI: We’re talking about generative AI, which excludes other types of AI such as Alexa/Siri, transcription services and digital games. Companies in the generative AI space include OpenAI, Anthropic, Google, Microsoft and Perplexity.

We also mean the public-facing chatbots that connect with generative AI systems operating in the background: Claude, ChatGPT and Copilot for Anthropic, OpenAI and Microsoft, respectively. Perplexity and Google’s Gemini use the same name for both their system and chatbot.

Choosing between chatbots is like picking between Dairy Queen and Baskin-Robbins; either will sell you an ice cream. But that decision gets you only to the door. From there, you need to choose a flavor — in AI-speak, a model. Each chatbot offers several models, so matters complicate quickly.

Since everyone’s preferences are different, learn the model you prefer … through a taste test.

Summer is the perfect time for this: Devote a week or two to exploring each chatbot, learning its plusses and minuses through your own prompts. (Upgrading to a one-month paid account will give the fullest insights.) This tryout will help you find a model suited to your needs.

The last bottlenecks

While Silicon Valley and Wall Street enthuse constantly about generative AI’s potential, don’t conflate commercial interests with the needs of schools.

One of the purported opportunities of AI, for example, is how it improves efficiency — doing more with less effort.

The subtext is often cutting costs, a siren song not just to business investors but also to schools with tight budgets. If AI does drive job cuts in education, the technology also makes communications especially vulnerable.

Essayist Ben Thompson has pointed out how tech removed many of the constraints in mass communications. Those remaining were at the start of the process, “the creation and substantiation of an idea” — in a word, storytelling.

Writing two months before the public release of ChatGPT took generative AI mainstream, Thompson aptly forecast how large-language models would eliminate these final bottlenecks, making content creation cheap and easy.

However, he didn’t anticipate the degree of backlash, chiefly among younger generations, against what we now know as “AI slop.” Despite its frequent use of generative AI, Generation Z is particularly discontented by the technology.

Communicators have an opportunity in this moment to showcase their value to schools. Commit to the kind of content that AI can’t replicate — campus-based stories that 1) illustrate a school’s values, and 2) rely on the kind of cultural knowledge that an algorithm would never have.

But comms does not face an all-or-nothing situation. Enlist AI for general tasks where chatbot copy would be perfectly acceptable — any time a well-meaning intern’s work would suffice.

Not everything can be encoded

The Achilles’ heel of generative AI is this: What constrains a system’s ability is what it “knows,” the data it has collected from either users or the Internet.

Stern put it this way in her book: “No data for AI, no growth. No survival.”

Where information is missing, AI will predict and confidently provide an answer to fill in the gaps. Unfortunately, these made-up answers often aren’t accurate.

Independent schools are more likely to keep important details offline than are public schools or higher education. Data invisibility, then, is an issue for school leaders to anticipate before giving generative AI more responsibility.

Writer Nilay Patel recently described a common assumption in Silicon Valley: Every important contribution can be captured in a database. From there, the information will eventually “make things happen in the real world.”

This way of thinking created our modern world, Patel wrote. From the streaming services we watch and the computers in our vehicles to the email system that put this newsletter onto your screen, much of our lives both produce and run on data.

“Once you start seeing the world as a bunch of databases,” he continued, “it’s a small jump to feeling like you can control everything if you can just control the data.”

That explains why, Patel wrote, “getting everything in a database so software can see it is a preoccupation of the AI industry.”

This interest has reached schools, too. Basic questions from prospective parents go to AI-powered platforms, rather than to employees. Marketing automation now drives many admissions practices. A “data-driven mindset” has become a key qualification for communications directors.

Yet what attracts people to our schools, what inspires them to enroll and accept job offers and stay for many years, hasn’t changed in generations: It’s the human factor. The staff member who always knows a colleague’s name, the teacher who honored a student’s bravery, the effusive recommendation from a past parent or graduate.

Very few of these moments, these contributions, would be captured in a database. Even so, they are essential to our institutions being distinctive and valuable.

Refining what you have

Instead of embracing data obsession — what Patel calls “software brain” — and pushing for ever more digital information, schools would be better off using AI to understand the information we’ve already collected.

Take website analytics. These hold plenty of lessons and insights, although few communications offices spare the time to examine them. Turn the analysis over to AI: Download a batch of recent user behaviors, then have your chatbot of choice report on the notable takeaways.

Do the same with your bulk-email engagement reports. Do you get better results on a particular day of the week? In the morning or afternoon? AI analysis can tell you. The same is true with your social media engagements.

Have you ever analyzed your magazine mailing list? AI could help you find a previously unnoticed hotspot of alums in a certain region.

Competitive analysis is yet another task that few schools have the time to perform. Assign your chatbot to gather and report on the key marketing messages of neighboring schools.

Other departments in the school can perform similar examinations of the data already in their systems.

In spite of the buzz, this is still early days for AI. With the technology’s constantly evolving capabilities, and schools’ unique use cases, it’s not easy to build on the experience of others to design productive and time-saving assignments.

So spend the summer to advance your AI understanding. By the time classes resume in the fall, you’ll be better prepared to join school discussions, to gently note where hype has outrun reality and to anticipate future asks of your office.

Copyright © 2026 Fine Point Communications. All rights reserved.