top of page
  • Writer's pictureRonnie Ray

How will AI Affect the Future of Work?

A working paper published by the University of Pennsylvania in partnership with researchers from OpenAI, takes an early look at the potential impact of Large Language Models (LLMs) and complementary technologies and tooling built on top. The paper does not purport to define actual declines in employment due to higher productivity and automation driven by LLMs, since social, regulatory, economic and other factors will all have an yet undetermined effect. However, the research findings that up to 49% of US workers could have half or more of their tasks exposed to LLMs and its downstream applications, is significant enough to set organizational strategists and skilled professionals, to start thinking about how to navigate the future of work.

Applications of LLMs

The emergence of generative AI models, particularly LLMs, has brought about many potential applications across various industries. In addition to text generation, LLMs can also be used for tasks like custom search applications, summarization, and classification. While the deployment of LLMs as versatile building blocks for creating additional tools will require time and reconfiguration of existing processes across various industries - this process is well under way.

LLMs are increasingly being integrated into specialized applications in fields like writing assistance, coding, and legal research. In addition, the adoption of LLMs into domain specific workflows —including tooling, software, or human-in-the-loop systems—to address issues such as factual inaccuracies, inherent biases, privacy concerns, and disinformation risks - is already in progress. The study cites the example of Casetext, which offers LLM-based legal research tools that provide lawyers with quicker and more accurate legal research results, utilizing embeddings and summarization to counter the risk that GPT-4 could provide inaccurate details about a legal case or set of documents.

Moreover, LLMs can become valuable assets in machine learning model development, serving as coding assistants for researchers, data labeling services, or synthetic data generators. Additionally, there is potential for such models to contribute to economic decision-making at the task level, refining methods for task and sub-task allocation between humans and machines. For example, one could imagine the tasks of online promotion and search engine optimization, segregated into idea generation by human critical thinkers, but actual execution programmed through machine models with active feedback loops directing automatic changes to future content.

Impact on Labor Markets

The following highlights showcase the pervasive nature of the impact of AI on current job occupations -

  • 19% of jobs have at least 50% of their tasks exposed to LLMs when considering both current model capabilities and anticipated LLM-powered software.

  • Accounting for other generative models and complementary technologies, up to 49% of workers could have half or more of their tasks exposed to LLMs.

  • Most occupations exhibit some degree of exposure to LLMs, with varying exposure levels across different types of work, but occupations with higher wages generally present with higher exposure.

  • Science and critical thinking-heavy roles show a negative correlation with exposure, while programming and writing skills are positively associated with LLM exposure.

Caveats and Actions

While the study presents data from human annotators knowledgeable about LLMs, there may be an element of subjective bias in the task breakdown and labeling used to assess exposure, since the researchers were not domain experts in every occupation that was analyzed. Also, the current analysis applies mostly to US job classifications, and further research is needed to address the impact on the workforce, and the way work is performed in different regions or economies. Additionally, future evolving capabilities and regulation may accelerate or decelerate some of the impact.

However, as the authors state, as LLMs have consistently improved in capabilities over time, their growing economic effect is expected to persist and increase even if the development of new capabilities were halted today. In short, some of these impacts may surface earlier or later, but the direction is set. It is therefore essential to continue monitoring the impact on the workforce and prepare for the inevitable changes to the future of work.


Subscribe and stay up to date on news from Trumio! 

bottom of page