Hardly a day goes by without artificial intelligence in the news.

On the positive side, AI optimists extol recent medical breakthroughs, such as the development of a drug candidate to treat pulmonary fibrosis, and the use of AI to surface unreported GLP-1 side effects in Reddit posts. They celebrate outstanding achievements in mathematics, like OpenAI’s solving of the Navier–Stokes equations, once considered unsolvable.

On the downside, skeptics worry about mass layoffs as companies increasingly adopt AI and the elimination of entire swathes of entry-level jobs from the labor market. Others note with distress the adverse environmental impact of AI data centers.

Most recently, powerful voices from within the AI community — the same industry leaders who create frontier models like ChatGPT, Claude, and Grok — have raised the alarm. They argue that recursive self-improving AI (RSI), without guardrails, could enable dangerous misuse and erode human control over the technology. This so-called X-risk scenario could become reality when models reach superintelligence, and Artificial General Intelligence (AGI) is achieved.

Both the signal and the noise surrounding the artificial intelligence debate are of particular
importance in the field of education. With previous technological innovations, like the
introduction of calculators, broadband Internet access, and 1:1 student devices, schools were able to plan ahead, field test and buy education technology tools while training teachers for their implementation in the classroom. All of these were adoption technologies.

Generative AI (GenAI) is different. It is, in the words of MIT Professor Justin Reich, an “arrival technology,” having landed uninvited in schools, with students and teachers starting to use AI tools after ChatGPT was released in November 2022.

Over the last couple of years, the ConVal School District has been on a deliberate journey
toward the introduction of GenAI. After adopting a School Board policy governing the use of GenAI, the district spent the last year and a half developing a Generative AI Use Plan, which first builds AI literacy and then moves to the gradual adoption of approved tools. The district website hosts the plan.

The core objective of the Use Plan is the safe, ethical, efficient and sustainable adoption of AI for specific use cases. The Plan puts the concept of the Human in the Loop (HITL) at the center of each AI decision, including the preemptive choice whether to use AI at all. Users actively prompt, guide, review and approve an AI tool’s product to improve accuracy, control for possible bias and verify alignment with standards and pedagogical goals. With AI, consistent human oversight ensures users never abandon their expertise and, at each step, remain in charge of the final output.

Once teachers have successfully integrated GenAI tools into instruction and assessment, the next step is to prepare students for their future with AI, with the goal of having them use it judiciously. Between the promises of frontier model breakthroughs and the perils of some AI companies’ unchecked ambitions lies the middle path of prudent preparation for the future. ConVal’s approach to adopting GenAI seeks to safeguard intentional use of the technology in education, one Human-in-the-Loop decision at a time.

Helfried C. Zrzavy is ConVal’s technology integration specialist.