What if we teach what artificial intelligence (AI) is before we dive in? Hollands & Breazeal (2024) reviewed the MIT Day of AI initiative and its use in the 2022-2023 school year. The survey shows that more than 50% of the students and teachers, using materials from the Day of AI, learned about AI, its effects on society, and places where students could use it ethically. In other words, the materials provided gave teachers and students a structure by which to learn about AI directly and how to use it. The trick is not to jump in the deep end, but ease into it.

I recently visited my mom, who is in her 90s and very active and engaged. She asked me what AI was. She had no real idea of what it was, how it was used, how she was currently using it on Google (or that she was even using it). She didn’t know if it made things better or worse. I tried to explain it verbally. No luck. She could not grasp the speed of the answers, which is one of the most powerful uses of AI. So I grabbed my computer and pulled up ChatGPT (since she had heard about that one). I showed her how to ask ChatGPT for a presentation title on a subject. Then I prompted ChatGPT to add humor in the title. Finally, I asked it to present me with 10 options with titles under eight words.

Mom was stunned to watch how quickly the basic stuff appeared on the screen. Then, she asked why I bother with AI when I could develop a presentation title independently. I can, but this is a fast way to start with options. I rarely take the title directly from AI, but sometimes it gives ideas or suggestions that are part of the final product. When mentally reviewing this discussion, though, I realized that my mom got to the heart of the matter – why bother with AI when I can do the task without it?

AI can help start or speed up a process. It can gather information at tremendous speeds. Teaching about AI is more than telling students (or adults) about what can go wrong. It is about understanding how you can use it in daily life. Touretzky et al. (2019) listed five ideas that K-12 students should learn (pp. 9797-9798):

1. “Computers perceive the world using sensors.” Sensors are items that bring data into the system. For example, Siri and Alexa are voice agents. Data is also gathered from keyboards, webcams, and internet connections that provide additional data like location and computer identification information. Students need to learn how to use different inputs, like voice agents. It also includes understanding the limitations of the sensors and when computers get it wrong.
2. “(Computer) Agents maintain models/representations of the world and use them for reasoning.” Students should understand that a map is a representation of a territory. Models can mimic real-world items, but they have limitations. For example, can your high school students create a word set that allows them to fully describe something? After they make a list of words, do they find themselves using words they didn’t add to the word set? This example shows students just how difficult it is to be completely inclusive. And, what happens if AI is not completely inclusive? What if AI is missing some words from its word set?
3. “Computers can learn from data.” This is the key to the growth of AI over the past several years. Students can “teach” computers to recognize hand signs. What goes into this teaching? What happens when the AI misunderstands? How easy is it to correct the misunderstandings?
4. “Making (computer) agents interact comfortably with humans is a substantial challenge for AI developers.” AI needs to be close, but not too close, to be useful. Does AI hear a human muttering as a command, or just something that should be ignored? Are the words AI uses creating a positive or negative environment? Are they used correctly?
5. “AI applications can impact society in both positive and negative ways.” Can students identify how AI is helping them? Can they identify the negative impacts? What about the potential problems? Is AI transparent enough that you can check how it came to the answer it got? We require this of our students; why wouldn’t we require it of our systems? Students need to be capable of researching the results to verify its accuracy.

I teach high school students. I am the last stop before adulthood, and I often consider what they need to know before they head out into the adult world. Learning to use artificial intelligence is not a simple, straightforward process. It is an iterative process in which we ask questions, get answers, double-check responses, and move forward. For our students, teaching about AI means that they can see themselves in the future, using and programming the next generation.

Plus, I show students what happens when AI gets it wrong. AI can hallucinate in funny and terrifying ways. AI can provide summaries of biased data to employers. AI can respond quickly. And, AI needs to be reviewed and verified, which may not be a quick process. This may be the most important lesson I teach.

References

Garg, S. (2023). Women May Pay a “MOM PENALTY” When AI Is Used in Hiring, New Research Suggests. Tandon School of Engineering, New York University, December 12, 2023. https://engineering.nyu.edu/news/women-may-pay-mom-penalty-when-ai-used-hiring-new-research-suggests

Hollands, F., & Breazeal, C. (2024). Establishing AI Literacy Before Adopting AI. Science Teacher, 91(2), 35–42. https://doi.org/10.1080/00368555.2024.2308316

Touretzky, D., Gardner-McCune, C., Martin, F., & Seehorn, D. (2019). Envisioning AI for K-12: What Should Every Child Know about AI?. Proceedings of the AAAI Conference on Artificial Intelligence, 33(01), 9795-9799. https://doi.org/10.1609/aaai.v33i01.33019795

Weiser, B. (2023). A Man Sued Avianca Airlines. The New York Times, May 27, 2023. https://www.nytimes.com/2023/05/27/nyregion/avianca-airline-lawsuit-chatgpt.html