Will AI take your job? OpenAI has the answer!

GPT-4, the latest AI and deep learning milestone, is capable of doing extraordinarily well on a wide range of professional and academic fronts. It, for example, passes a simulated bar exam and ranks in the top 10%. It also passed some of the most challenging exams, including the LSATs, SATs, and GRE. Large Language Models (LLMs) are improving at doing human activities. With such astounding findings, everyone is apprehensive about whether AI will replace them.

OpenAI, OpenReseach, and the University of Pennsylvania published a paper, "GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Model", to analyze the potential impact of such tools on the labor market. In this article, we'll look closer at their findings and discover which jobs are most likely to be harmed by the growth of LLM.

Summary

• According to the study, at least 50% of US jobs can be affected by LLMs.
• 10% of tasks in 80% of US workers can be done by LLMs.
• For the other 19% of workers, LLMs could influence at least 50% of tasks.
• Higher-income jobs will potentially face greater exposure. Programming and writing jobs will also have an impact.
• Jobs that require critical thinking and science are safe. Similarly, jobs with a low barrier to entry are less likely to be impacted.

How they conducted the research

Researchers used O*NET 27.2 database. It is the database of occupations and employment in the US. It comprises 1,016 occupations with their Detailed Work Activities(DWA)s and tasks. And then, they used a rubric to collect humans and GPT-4 generated annotations to decide whether LLM can cut the time required for a human to perform a certain task by at least half. One thing to note here is that LLM, in this case, refers to GPT-4 or secondary systems built with GPT-4.

The results are based on the exposure rubric. Exposure measures whether access to GPT or GPT-based systems would reduce the time required to perform a specific DWA or task by half. The more exposure, the more likely LLMs will impact the job.

You can read the entire paper here.

The results of the study

According to the findings, LLMs are more likely to have an impact on jobs with high salaries, mainly programming, and writing. Jobs that demand critical thinking and science, on the other hand, are less likely to be harmed.

Another observation is that LLM is more likely to overtake jobs with high entry barriers. Individuals holding Bachelor's, Master's, and similar professional degrees are more exposed to LLMs than those without formal education.

Here is the list of jobs with the highest exposure:

• mathematicians
• tax preparers
• writers
• web designers, programmers
• accountants
• journalists
• legal secretaries

Jobs that are less likely to be impacted by GPT are as follows:

• graphic designers
• search marketing strategies
• financial managers

The researchers also list LLM's impact on various industries. Industries with the most significant impact are as follows:

• data processing services
• information services
• publishing industries
• insurance carriers

On the other hand, the industries with the most negligible impact are:

• food manufacturing
• wood product manufacturing
• support manufacturing
• agriculture forestry

Conclusion

The study compares Generative Pre-trained Transformers with General Purpose Technologies as its conclusion. Researchers conclude that as GPTs such as fire, electricity, and mobility significantly impacted the economy, neural networks, LLM models, GPT, and GPT-based systems will substantially affect the world. They hypothesize that GPTs(LLM Models) would have the same impact on the world as GPTs(General purpose technologies).

It won't be long before AI is used regularly in the workplace as Large Language Models progress. It can provide even better work in less time with fewer hallucinations, greater accuracy, and consistency.