Will OpenAI's ChatGPT replace your job? A new study from the company estimates that AI-powered chat technologies could seriously affect 19% of the jobs in the US.
The AI-powered ChatGPT chatbot is powerful enough to write essays and marketing pitches, program computer code, and extract insights from financial reports, and OpenAI researchers estimate that ChatGPT and future software tools built with the program could impact at least 50% of the tasks necessary for around 19% of the jobs in the US.
Meanwhile, 80% of the US workforce could see at least 10% of their work tasks affected in some way by ChatGPT, which was recently upgraded with a new GPT-4 model.
The study adds: “Our analysis indicates that the impacts of LLMs (large-language models) like GPT-4, are likely to be pervasive.” In addition, researchers found that jobs with higher wages—which can involve the worker performing many software-based tasks—could face more exposure to potential disruption from AI-powered chatbots.
“We discover that roles heavily reliant on science and critical thinking skills show a negative correlation with exposure, while programming and writing skills are positively associated with LLM exposure,” the study says.
OpenAI researchers cataloged which professions could see the most disruption using various measurement rubrics. The most affected professions included interpreters and translators, poets, lyricists and creative writers, public relations specialists, writers and authors, mathematicians, tax preparers, blockchain engineers, accountants and auditors, along with journalists.
The paper also breaks down the ChatGPT impact by industry. Sectors including data processing hosting, publishing industries, and security commodity contracts, saw the most potential exposure to disruption. In contrast, industries known for manual labor—food services, forestry and logging, social assistance, and food manufacturing—saw the least potential impact.
That said, the study has several limitations. OpenAI looked at the over 1,000 professions in the US, and labeled them with various tasks needed to perform the jobs. Researchers then used human annotators and a GPT-4 model to rate whether access to a ChatGPT-powered system would cut down the time required for a human to perform a specific task “by at least 50%.”
Hence, the study itself concedes there’s an inherent bias in trying to sum up each profession by using simple labels to describe job tasks. The study adds: “It is unclear to what extent occupations can be entirely broken down into tasks, and whether this approach systematically omits certain categories of skills or tasks that are tacitly required for competent performance of a job.”
The other issue is that GPT has shown it can make obvious mistakes, including making up information, which makes it necessary for a human to oversee the work. That’s a factor the study wasn’t able to take into account. In addition, the study only looked at whether ChatGPT could reduce the amount of time needed to complete various tasks by profession. This doesn't mean ChatGPT is necessarily smart enough to fully automate certain jobs.
Still, the researchers anticipate ChatGPT and its future iterations will shake up the way people work. Thus, society and policy makers need to prepare. “While LLMs have consistently improved in capabilities over time, their growing economic effect is expected to persist and increase even if we halt the development of new capabilities today,” the paper adds.