As AI invades, human intelligence becomes more important than ever in schools | Opinion

Dr. Thomas C. Hammond

Anyone who regularly uses network technology, at this point, is familiar with the phenomenon of AI invasion. AI features have been added to software that has been used for many years.

Since 2011, the iPhone has encouraged users to choose whether to use Siri or not. Recently, Android phones are trying to use Google's Gemini. If you are using Microsoft products, Copilot is included. Opening a document in Adobe Reader immediately provides an AI-generated summary.

The same trend is being carried out in schools. For example, many teachers and students use Padlet, a virtual message board that many schools purchased during the community pandemic. Padlett is now integrated with Adobe's Firefly, allowing students to create and share images generated by AI. The final result creates both opportunities and challenges. Teachers and students can use this new AI integration to enhance learning, engage in distracting or disruptive behavior, and even both.

Like everything that includes technology, the answer is “both.” After all, hammers can be used to build and destroy hammers, and all the rules and safety procedures introduced by shop class teachers can reduce counterproductive applications rather than eliminate them.

I am fascinated and scared of AI that can be used to learn.

As a high school teacher, technology director, and (now) professor of education technology, I am a 21st century shop class teacher, and AI is the latest technology to participate in the curriculum.

However, AI is a very different kind of tool than a hammer. The hammer does one (or several). AI does a lot. Organizing which tasks are best done through AI and which tasks aren't being done can be much more difficult. (Try using it to create a crossword puzzle. For this relatively simple task, it's currently awful.)

Many AI users stopped Google searches and replaced them with questions to Gemini instead. This is a trend that Google encourages by placing AI-generated results at the top of every page in a search response. Another problem is that the quality of AI output is difficult to determine, requiring both contextual knowledge and vigilance.

Try this experiment. Ask AI for images of LenapeBillage. Specify a year, if necessary, 1400, 1900, or even today. The images are historically wrong. This is wrong about modern life in Lenape, including inaccurate houses in the wild, canoes, and planting methods.

Now, try another experiment. Ask AI Tools for stories about a LENAPE person's life day. Again, specify 1400, 1900, or today's year. The story is much more accurate than the images! Why is this? You can't give an accurate answer. I don't know if AI developers can do it. However, I think it has something to do with the difference between the training data used by the image generator (other images) and the training data (text) of the chatbot. History exists primarily in written form, and linguistic processing can produce plausible historical stories or reliable explanations of historical events.

On the other hand, image generation training data contains very little historical images. When asked to construct an image of Lenape Life, AI essentially grasps the straw…and users may not know this.

Why is this important to schools? First, like everyone else, students and teachers need to understand the pros and cons of AI as tools. Secondly, and unlike all others, teachers need to teach AI literacy by showing their strengths as well as their weaknesses.

AI is currently the weakest in visual and spatial tasks, such as creating images for Renape Village and crossword puzzles. By ensuring students can confirm that human intelligence is more powerful than AI, teachers can help them realize that AI is, after all, a tool that is not different to hammers. Human intelligence and judgment are required to exercise AI responsibly for productive purposes, and careless or malicious use can be very destructive.

And, just like infants swing a hammer, the destructive use of AI is much more achievable than productive.

Some students use AI productively, but many already use it to learn their creativity and plagiarize or take intellectual shortcuts that blunt. In response to this trend, some schools have chosen to retreat to the pre-learning world, restricting students' access to technology during their studies, requiring, for example, to handwritten all tasks.

To prepare more meaningful education, and after-school students is to understand the limitations of AI and know that human intelligence and judgment are important for its use and our future.

Dr. Thomas C. Hammond is Associate Professor and Director of Education, Learning and Technology Programs at Lehigh University College of Education.