Chatting in Tongues: a critical need for language models

by Marie Lena Tupot and Tim Stock, scenarioDNA inc.

The discourse of late surrounding ChatGPT and other AI tools makes us all sound like anxiety-laden neophytes. Many of us have been decoding the evolution of language for decades now. So why is this scary? We weren’t afraid of the Internet when it was in its infancy. We weren’t afraid of Natural Language Processing or discourse modeling. Nor were we afraid of the power of a Google search. Yet, here with AI we are stuck. Is it because we have lost the capacity to think beyond what we already know, to take risks? 

We should immediately be seeing AI as tools we need to sharpen, tools that help push our human capacities toward real innovation. 

Nick Cave’s ChatGPT takedown when someone from New Zealand dared to prompt ChatGPT to write a song in the style of Nick Cave is silly. Of course, the resulting song is a “mockery of what it is to be human.” The machine can only work from Nick Cave’s existing oeuvre. ChatGPT itself tells us more about who Nick Cave is ideologically than Nick Cave does. A simple prompt of “what is nick cave’s ideology?” explains “it is difficult to say if his views align with any specific ideologies.” Of course, it is difficult to align Nick Cave with an ideology. To be Nick Cave is to be divergent. To be divergent is to be human.

For the most divergent of us, we need these tools to take us even further. ChatGPT isn’t there yet for Nick Cave. It’s not there for cognitive linguists either. 

Even PeopleAI isn’t yet speaking to us in the tongue of the historic figure in question. It’s pulling from facts surrounding the figure like a dynamic wiki page. For example, it is impossible to get Joan of Arc to explain how she might be radical today and what she would fight for. Her chat simply keeps reiterating “I’d be radical in my advocacy for the marginalized and oppressed.” Aren’t many of us doing that already? It’s a good example of where ChatGPT is currently at.

What we should want to hear is Joan of Arc’s authentic 19-year-old voice, her rhetoric, speaking to us today from the past. First, we would need to decide who is the true Joan. Her legend is split into two ideologies. Allison Miller wrote for JStor at the centennial of Joan’s canonization, “Within France, she is a symbol of reactionary nationalism, venerated by the Far Right long before she was canonized. Outside France, though, Joan has been more of a heroine of feminism and androgyny, especially in Britain and the United States.”

These points could be AI executions that challenge our biases. Perhaps we might want to hear our own voices so we can check ourselves. Or maybe we need more antagonistic voices in our heads to take that one step further? This relies on building patterns of linguistics and archetypal coding. It requires archetypal frameworks for language. We established one such structure with our culture mapping patent in 2015.

Joe Pompeo writes for Vanity Fair in response to AI, “If you’re an analysis person, let’s say, 20 years down the road, you might need to find something else to do.” However, analysis requires seeing meaning in patterns, not simply seeing patterns. We’re safe, Joe. Meaning, like divergence, is a human concept.