Researchers from the Stanford Institute for Artificial Intelligence have developed an algorithm that can estimate salary levels from a job description. VentureBeat reports.
To map the relationship between the text of the posting and pay, the researchers assembled more than a million records of job searches. They then used a natural language processing model BERT and, with its aid, processed 800,000 vacancies.
During testing on the remaining records, the algorithm proved able to predict the match between the job description and salary with 87% accuracy. By contrast, using only job-posting headlines and geographic locations yielded accurate predictions in only 69% of cases.
Using the algorithm, the researchers also identified a link between possessing certain certificates and prospects for a pay rise. Possessing such information will help workers choose truly necessary upskilling courses, they say.
Lead researcher Sara Bana noted that not only job seekers will benefit from the algorithm. In her words, employers can use this information to invest more effectively in human capital and retrain certain employees in advance.
The authorities will use this data to promote professional education programmes and understand the economic value of skills, Bana added.
“Such tools will help job seekers improve their employment prospects, employers — the workforce, and the authorities respond to immediate changes in the economy,” said the researcher.
In the future, Bana hopes to develop a web application in which workers or companies can explore the added value from upskilling or relocating to a new locale.
The scientist will also attempt to characterise the contribution of different words to the salary forecast.
“Ideally we will colour-code words in postings from red to green, where dark red words are associated with low salaries, and dark green with high salaries,” she says.
In May, Singapore authorities announced mass deployment of robot assistants by local businesses due to labour shortages.
In June 2021, researchers found that automation and artificial intelligence have contributed to wage declines of 50-70% in the United States over the past 40 years.
Follow ForkLog AI on TikTok!