LLMs Will Change Jobs, Not Just Tasks.

Credits: Calum Heath

There is no doubt that the rise of Artificial Intelligence (AI) tools is poised to have a profound impact on the job market. A recent New York Times article, "The Worst Part of Wall Street Career May Be Coming to an End," argues that AI is set to replace a significant portion of entry-level white-collar work on Wall Street. This signifies a shift beyond just automating specific tasks; AI has the potential to fundamentally change the types of jobs available.

This aligns with the concept of Large Language Models (LLMs) impacting entire jobs, not just individual tasks within them. Ethan Mollick, a prolific writer, an expert on LLMs, and a professor at the Wharton School, University of Pennsylvania, explains in his book «Co-intelligence. Living and working with AI» that there are three types of tasks:

• "Just me tasks" are tasks that only I can do and where the LLM can directly assist or replace the human's work, such as writing and composing.
• "Delegated tasks" are tasks where a human needs to check the LLM’s output and make the final call like editing a text.
• "Automated tasks" are tasks where the LLM can do a better job at finding facts, summarizing information, coding, or handling routine communications.

This raises an interesting question: How do we build expertise in a world where LLMs are increasingly capable of handling a wide range of tasks? Mollick argues that expertise is a function of several factors, including basic knowledge, active working memory, the ability to connect historical dots, deliberate practice, and having a mentor who can provide feedback.

The emergence of LLMs as potential "mentors" available at all times rather than by appointment, is an intriguing prospect. However, it is crucial to remember that LLMs are ultimately prediction machines, guessing the next word based on context. Consequently, several tasks still require human judgment for important decisions and to avoid potential inaccuracies or plain errors. The implication is that expert input will be in high demand.

As AI continues to advance, the impact on jobs and the job market will be significant. Rather than fearing the rise of LLMs, we should explore how to best harness their capabilities while maintaining the essential role of human expertise and decision-making. By understanding the different types of tasks and the factors that contribute to building expertise, we can adapt and thrive in this rapidly evolving landscape.