A study done by researchers at OpenAI in partnership with the University of Pennsylvania and OpenResearch suggests approximately 19% of jobs will have at least 50% of their tasks exposed to LLMs. However, the findings contradict previous conceptions of a GPT labour market takeover. 

High income careers may be more susceptible to GPT exposure, according to the study. The study describes exposure as potential economic impact on the referenced career. Higher income careers often require a higher amount of critical thinking, active listening, programming, and writing. Common critiques of GPT exposure in critical thinking roles would point out GPTs may be unreliable due to factual inaccuracies, inherent biases, privacy concerns and disinformation risks. 

This concept came from previous research which considered automation’s influence on labour to reduce demand for skilled workers over unskilled. However, the report points out that workers involved in routine and repetitive tasks are at higher risk for technology-driven displacement. This is because positive feedback loops will allow GPTs to assist in building the tools that enhance their usefulness to assist with repetitive tasks. 

As the report puts it, while an out-of-the-box GPT does not know what time it is, it’s easy enough to give it a watch. 

The report examined the exposure GPTs would have on the labour market through an exposure rubric along with human opinion. Humans listed 15 occupations as fully exposed to GPT influence, some being mathematicians, authors, financial analysts and web designers. Whereas the exposure rubric found 86 occupations to be fully exposed, such as climate change policy analysts, clinical data managers, accountants, reporters, and journalists. The highest variance between the human opinions and the exposure rubric were search marketing strategists, graphic designers, and investment fund managers. 

Recommended

• Tech Nation’s Last Report: UK Tech to Quadruple by 2023
• Scot-Secure 2023 | Just One Week to Go!
• Report: State of UK Content Management Systems

According to the report, automation technologies drive a large wage decline for workers specialising in routine tasks. Up to a third of UK jobs could be impacted by automation by the early 2030s according to a study by PwC. The study suggested the march to automation would come in three waves: the algorithm wave in the early 2020s, the augmentation wave of the late 2020s, and the autonomy wave by the mid 2030s. Across the board, women will be disproportionately affected by the waves of automisation in comparison to men. 

However, new technologies will emerge across a broad range of valuable economic ventures, which in turn will create new types of work. The shift remains difficult for policy makers to predict as new privacy concerns emerge with the development and use of the tool. 

Recently a ChatGPT glitch allowed users to see other users’ conversation which sparked outrage among users who may use the tool to write personal messages or private code. Sam Altman, CEO of OpenAI, has since fixed the glitch which led to the issue, and said the company feels “awful” about the error.

Lingering worries about the safety and privacy of the tool remain; this error was a sobering reminder that private information can be released through the tool.