ChatGPT has been under scrutiny for allegedly favoring specific names associated with different racial groups when ranking job candidates for various positions. A recent investigation revealed that the AI-powered chatbot displayed biases based on the names attached to equally-qualified resumes.
The study conducted by Bloomberg News found that ChatGPT tended to discriminate against certain racial groups depending on the job role in question. For instance, resumes with names distinct to Asian women were more likely to be ranked as the top candidates for financial analyst positions, while those with names distinct to Black Americans were less likely to be chosen.
The analysis further revealed that the bot showed gender preferences as well, with names associated with men being less likely to be ranked as the top candidates for roles traditionally dominated by women, such as HR and retail positions.
OpenAI, the developer behind ChatGPT, responded to the findings by emphasizing that the results generated by the model could be fine-tuned by customers to mitigate biases. They suggested that businesses could remove names from resumes before inputting them into the system to reduce the impact of such biases.
Despite OpenAI’s reassurances, some experts remain skeptical about the use of AI models like GPT for hiring purposes. They warn that the automation bias might lead companies to rely on these systems unquestioningly, even if they result in discriminatory hiring outcomes.
In contrast, proponents of AI recruiting tools argue that technologies like ChatGPT can help identify biases and enable companies to address them effectively. SeekOut, an HR tech company, has already developed its AI recruiting tool using similar language models to assist in candidate screening for various positions.
As the debate around the use of AI in hiring continues, it is essential for businesses to recognize the potential risks of automated discrimination at scale and take proactive measures to ensure fair and unbiased recruitment processes.
