How Task Skill and Worker Proficiency Interact with GenAI

Daily ChatGPT users grasp how profoundly this tool has become an extension of thought. Unlike any preceding technology, Large Language Models interface so fluidly they reshape cognition itself. Our ability to “sense,” "think," "learn," "decide" and "act" no longer reside solely within biological minds but rather intertwine with this external artifact.

ChatGPT and similar tools can significantly alter workflows by changing how we match tasks with skills. Think of a two-by-two matrix: on one axis, you have the skill needed for a task; on the other, the worker's proficiency level.

In the top-right quadrant, where both skill and proficiency are high, generative AI can amplify a professional's capabilities. Here, the individual is in a state of 'flow', working effortlessly with the AI. For instance, a proficient writer using AI as an editor can create better content more efficiently, with a seamless integration of AI assistance and personal expertise. This synergy minimizes errors and maximizes output quality.

When a high-skill task meets a less proficient worker, AI becomes a tool for bootstrapping performance, albeit with a higher error rate. In this scenario, represented in the bottom-right quadrant of our matrix, the focus should be on learning and skill development, rather than merely substituting for higher skill. For instance, a novice programmer using ChatGPT to learn coding is beneficial, but relying on it to replace an experienced programmer poses risks due to potential undetected errors.

In situations where the task's skill requirement is low, but the worker's proficiency is high (top-left quadrant), AI acts as a timesaver. An example is an expert writer using AI to quickly draft a bullet point summary of their essay. Here, the professional’s ability to check AI’s output remains crucial, though the summarization task itself is less demanding and time-consuming.

In the lower-left quadrant of our matrix, where both skill and proficiency are low, AI serves as a prosthetic. For example, a writer creating a simple social media post would find AI extremely useful for saving time and cognitive effort.

Across all these scenarios, the crucial element is human judgment. Regardless of skill level or task complexity, the use of generative AI highlights a shift: it decouples raw knowledge and skill from judgment, enabling us to apply our cognitive abilities more strategically and effectively.

This requires reevaluating notions of individual autonomy and intellect. If external tools constitute integral conduits for memory, creativity and task completion, they become part of our identity. Their capabilities shape our own.

To interact with Large Language Models is not to use mere software, but to enter an emergent cognitive system that reshapes the possibilities and limits of one's native capacities in their state of entanglement.

By reflecting on how "we" arrive at insights, judgments and ideas in tandem across biological and digital domains, we gain clearer perspective on the increasingly distributed nature of agency and intelligence in this age of cognitive outsourcing.