Nobody yet knows where large language models will ultimately prove to supply the greatest value, but it is fair to say observers believe a few use cases are good candidates for impact.
High-potential use cases include content creation, such as writing marketing copy. Customer service and communication supporting chatbots, answer queries, and personalized interactions are another use case viewed as having high and nearly-immediate value.
Code generation and software development is a third area of interest. LLMs can assist programmers with tasks like writing code, debugging, and suggesting improvements.
Many also believe LLMs will prove valuable for research, as LLMs can analyze vast datasets, generate hypotheses, and summarize complex information.
Some might argue that up to 60 percent or more of the total value produced by LLMs could come in those four areas.
But even in areas such as content creation, impact remains unclear. Many doubt LLMs can succeed for creative writing and art. But others might believe there is plenty of room for replacement of human effort.
While some might argue “great art or writing” might not be produced by LLMs, there is quite a lot of passable storytelling LLMs could generate. “Depth and context” might not be LLM's greatest strengths, compared to human creators.
But an argument can be made that some significant amounts of “interesting” content can ber created, though we might still debate the degree to which “originality” or “creativity” are demonstrated.
Still, LLMs ought to be able to develop new arrangements of established patterns in storytelling, for example.
For example, storytelling often is said to involve only about six or seven key plots. But as with musical chords, though there are only a few basic notes and chords, countless arrangements, melodies and harmonies can be built from a limited number of basic building blocks.
Christopher Booker's Seven Basic Plots include:
Overcoming the Monster: The protagonist confronts and defeats a villainous force threatening them or their world (e.g., Star Wars, Harry Potter).
Rags to Riches: The protagonist rises from poverty or obscurity to achieve great success (e.g., Cinderella, The Great Gatsby).
The Quest: The protagonist undertakes a journey to acquire an object or achieve a goal (e.g., The Lord of the Rings, Moby Dick).
Voyage and Return: The protagonist embarks on a physical or metaphorical journey, then returns with newfound knowledge or understanding (e.g., The Odyssey, Alice in Wonderland).
Comedy: The protagonist uses humor and wit to overcome obstacles and achieve personal or societal change (e.g., Almost Famous, Shaun of the Dead).
Tragedy: The protagonist's noble aspirations lead to suffering and downfall (e.g., Hamlet, Macbeth).
Rebirth: The protagonist undergoes a profound transformation, often involving death and renewal (e.g., The Lion King, The Matrix).
A 2016 study by researchers at Northeastern University identified six "core trajectories" based on a statistical analysis of narrative emotional arcs:
Rags to Riches: A rise in happiness and well-being.
Tragedy: A fall in happiness and well-being.
Man in a Hole: Fall followed by a rise.
Icarus: Rise followed by a fall.
Cinderella: Rise-fall-rise: A cycle of success and setback.
Oedipus: Fall-rise-fall: A complex arc with multiple reversals of fortune.
The concern many in the screenwriting, acting and other “creative” areas attests to the fear LLMs and AI in general might be capable of displacing much human effort and output.
We might be a century away from full AI replacement of human functions. But the concern seems well placed enough.
