AI Historical Avatars: Opening Doors to History or Spreading Misinformation?

AI Historical Avatars: Opening Doors to History or Spreading Misinformation?

Artificial intelligence (AI) is revolutionizing how we interact with the past, offering unprecedented ways to connect with historical figures. But alongside this exciting potential, there are real concerns about accuracy and the spread of misinformation. Projects like the museum's AI-generated Vincent Van Gogh avatar exhibit demonstrate the possibilities, as seen in this YouTube video:

The New York Times writes about the exhibit's potential and how these avatars will bring the past to life, giving us a more personal and engaging connection to historical figures (archived article). A recent CBS spot lite also excellent covers the exhibit found here.

Tools like Humy take this concept to classrooms (free as of February 10, 2024), allowing students to interact with historic figures in a chat-like format. Teachers can customize these avatars with documents and information to tailor the experience to specific lessons and learning outcomes. It is exciting to think how this can further bring history to life by making the experience more accessible and immersive.

Concerns About Accuracy and Authority

However, when speaking with educators, a concern was raised about using AI for historical avatars. Primarily, there is the potential for the AI to be inaccurate and cause misinformation. These avatars rely on AI models trained on datasets that may be incomplete or contain biases. The AI framework, Learning Language Models (LLMs), is also known to become inaccurate overtime and therefore needs constant monitoring. These systems learn from massive datasets, and new information or even changes in how people use language can make their older understanding out of date. While developers attempt to minimize this, mistakes can still slip through.

Extending this concern, it was brought to my attention that students might treat the avatar as an unquestionable authority. If the avatar propagates misinformation, students might unknowingly adopt and internalize this incorrect knowledge. It raises the question: do we already see this effect in historical movies and TV shows, and is there research to support this concern? And to what extent?

The Teacher's Evolving Role

Assuming teachers recognize these limitations, what is their role? Should such tools be avoided due to potential risks, or should educators leverage them while fostering critical thinking in their students? The latter seems preferable. Just as a parent watching a movie with historical inaccuracies can facilitate a discussion with their children, teachers can guide students to dissect the information to determine what is factual and what is fabricated. I remember when schools were rapidly connecting to the internet; the same concern emerged about the need for students to utilize critical thinking when consuming information online – one could argue that this need is even more paramount today.

Transforming Challenges into Opportunities

AI historical avatars present a fantastic opportunity to capture students' attention. Their novelty and interactivity can ignite a curiosity and passion for history that traditional methods might struggle to achieve. The key is for educators to embrace a dual role:

Curators of Content: Teachers should carefully select which AI avatars to use and thoroughly vet the information they contain, supplementing them with reliable sources as needed.

Coaches of Critical Thinking: Students must be taught to approach these avatars not as ultimate authorities, but as tools that require evaluation and cross-referencing with other sources.

The Power of Discussion

The potential for inaccuracies isn't a reason to discard this technology, but rather a valuable context for teaching critical thinking and media literacy. Teachers can encourage students to ask questions like:

• Is this information consistent with what I've learned from other sources?
• Could the AI be misinterpreting certain historical documents or events?
• Where does the data this avatar is using come from, and might that data contain biases?

Conclusion

By taking a mindful approach, teachers can turn  potential pitfalls into powerful teaching moments. AI-generated historical avatars undoubtedly have the potential to make history education more accessible, personalized, and engaging. The fact that these tools have flaws makes the learning even more legitimate. When coupled with critical thinking and cross-verification, these avatars could usher in a new era of active, student-centered learning about the past.