In just six short months, ChatGPT has begun revolutionizing the way we work, continuously revealing its remarkable potential. As we witness these rapid advancements, it's crucial to reflect on the impact of this Large Language Model (LLM) technology on the labor market.
AI can complete half of all worker tasks.
Research indicates that approximately 50% of the U.S. workforce could have at least 60% of their work tasks completed by LLMs, along with any software built on top of them.
This finding comes from a pre-published research paper in March, involving OpenAI researchers (who can hardly be suspected of overestimating the impact of ChatGPT-like technology on the labor market).
Consider the implications of this finding: half of all employees could have the majority of their work completed more efficiently by machines, simply by leveraging LLMs and LLM-powered software.
In a consistent finding, approximately 50% of all worker tasks in the US could be completed significantly faster and with the same level of quality using LLMs and LLM-powered software.
The study does not mention that tasks are sometimes performed with levels of quality not just equivalent, but far superior to what each individual could achieve on their own, regardless of the time they devote to it.
However, the existence of large-scale technical feasibility of automation does not necessarily mean that it will actually happen.
Tasks that can be automated, will be automated.
For a task to be automated, once the technical feasibility is confirmed, certain additional conditions must be met:
Value-creation In the case of LLMs, the value proposition is clear. The marginal cost of running ChatGPT is negligible, while the marginal benefit in terms of productivity and creativity gains is significant. For example, automating legal tasks could reduce legal fees by 90% - a compelling reason for automation.
Lack of preference for human interaction In most cases, there is no preference for human intervention over machines for tasks that LLMs excel in. However, social norms and preferences may favor human intervention for certain tasks, such as in psychology or caregiving.
Considering these factors, what could prevent a task from being quickly automated?
Regulation Regulations, such as the AI Act in Europe, aim to uphold the standards of trustworthy AI. This includes ensuring human oversight, algorithm explainability and transparency, lack of bias, and data privacy. These regulations will slow down implementation, particularly in high-risk categories.
Adoption delays Minimal adoption delays are anticipated, as Generative AI is being adopted at breakneck speeds. ChatGPT reached 1 million users in just 5 days and hit the 100 million mark within a month. In contrast, it took approximately 40 years for electricity to permeate society after the introduction of electrical motors.
Value-creation is high; there is no preference for human interaction; adoption delays are minimal; regulation is not in place yet, although potentially limiting. As an overall result, it's clear that tasks which can be automated by LLMs will be automated - and quickly.
There will be an upheaval in the labor market
Automation will cause some jobs to disappear, while new jobs will emerge. In most cases, however, existing jobs will undergo drastic changes as tasks are automated, recombined, and job scopes evolve.
Different outcomes can be expected depending on the situation:
If your core tasks are automated, your job is at higher risk of being recombined with other jobs or disappearing altogether. This includes professions like translators, copy editors, developers, and analysts and many others.
If only peripheral tasks are automated, professionals will experience increased productivity. For example, strategists can use LLMs to augment their work significantly, at least initially.
In each case, the main unknown in determining the impact on employment remains the resulting demand.
Highly demanded and currently scarce jobs, such as doctors and teachers, will persist despite the automation of many time-consuming tasks. Increased productivity will allow them to focus on higher-value tasks.
Even for highly automated tasks, such as generating marketing campaigns, it is difficult to know whether many more will be generated with the current human workforce, or if the workforce will decrease significantly due to the boosted productivity for a demand that hasn't increased in the same proportions.
How one should prepare for it
With the upcoming upheaval on the job market, the rise of AI does have 3 major implications for everyone:
Managing change At the very least, your job will evolve significantly due to automation of some of its components. You need to embrace change and excel at adapting accordingly.
Using tech If you do not use this newly available technology, someone else using it will quickly replace you. Make sure you know how to use tech.
Honing human skills The way to shield and differentiate from machines is to excel at those skills that machines cannot easily replicate Focus on socio-emotional skills, critical thinking, creativity and other skills deemed typically human.
This is a reminder that everyone should embark on a continuous upskilling journey to prepare for the future of work.
We at Elendi can help: it is our mission and our expertise.