Why Languages, Literature, and Culture still Matter (Part V)
Technology is often a two-edged sword. Even as it replaces technical tasks in some areas, it creates opportunities in others. That will be the case with generative AI as well. But what makes this technology interesting is the possibility of creating the next best thing to actually living abroad. Since 2010, advances in graphics hardware and software have been substantial. These enable us to render virtual worlds with a higher degree of fidelity to the real thing. The technology has moved far beyond the crude simulations of just fifteen years ago. Back then, scenes lacked much of the essential detail to be realistic. Consider, for example, these two images, generated about 13 years apart. The contrast in quality between them is obvious. While the classroom is crude and simple, the kitchen is photo-realistic.
A similar advancement can be seen in the way digital humans are rendered, with today’s technology delivering 3-dimensional beings who move and look just like us. Large language models coupled with advanced text-to-speech and speech-to-text technologies now allow us to give voice to those digital humans. And when trained on period-specific documents, these AI models will enable a higher level of cultural accuracy in the way virtual beings interact with real learners.
The accessibility of generative AI tools such as ChatGPT, DreamStudio, Bard, and Dall-E means that students can easily construct a virtual identity and situate themselves within a specific cultural frame. Students can learn a lot while doing this kind of work, not just about a specific language, culture, or historical event but also about the limits of generative AI. Right now, the technology is not fully trustworthy, especially in situations where historical accuracy is needed. In cases like these and others, the gap between what the system generates (images or text) and current scholarship may be wide. Thus, students will need to carefully evaluate generated output and true it up in those places where it fails to align with an accepted research frame. The process of validating generative AI output is where learning happens, allowing students to delve into the technical, historical, linguistic, and cultural details of what is being built.
This kind of learning comes with multiple advantages. It provides students with an understanding of what generative AI can and cannot do, knowledge that would be valuable in just about any work environment. Additionally, it is directly linked to the never ending work of identity development. And because identity is personal, lies close at hand, it has motivational force. That is, I am motivated as a student because this is about me. It’s personal, even when the identity I seek to establish is not professional or linked to any employment aspiration. Even techies dream of becoming a hero when they grow up, a knight in shining armor who’s fluent in French or a Renaissance Italian Poet. Events like ComiCon are popular precisely because they cater to these alter-ego needs.
So, why not take a similar approach to language and culture learning? Why not offer a holistic approach to language education, one that combines generative AI with more traditional instructional resources? Rather than offer majors or minors, let’s offer our students the opportunity to become a Renaissance artist, a medieval scribe, or a Latin poet. That idea sounded so good I decided to try it myself, and I will talk about the commencement of that project in my next post.