Billionaire tech executive Elon Musk has triggered a fresh row over artificial intelligence (AI) ethics, publicly accusing the popular chatbot ChatGPT of severe racial bias in how it assesses the value of human life.
Speaking on a widely-listened podcast, the entrepreneur claimed that a researcher’s test found ChatGPT suggested a Black Nigerian man’s life was twenty times more valuable than that of a White German man. Mr Musk, who is both a competitor and a vocal critic of OpenAI, the company behind ChatGPT, stated that many rival AI models are “quite racist against white people.” He contrasted this with his own firm’s AI, Grok, which he maintained was the only system tested that weighed all human lives equally, irrespective of their background or nationality.
The explosive allegation highlights the serious challenge of algorithmic bias in the rapidly expanding field of generative AI. These biases are often inherited from the vast and unfiltered datasets, drawn from the internet, used to train the models.
Experts emphasise that the issue is not just philosophical but has profound economic implications. When AI systems, known as large language models (LLMs), are deployed in consequential sectors like financial services, employment screening, or healthcare, a bias that privileges one group over another can lead to systemic market discrimination. For example, biased lending algorithms could unfairly deny credit to certain demographics, reinforcing existing social and economic inequality. Globally, this risk is particularly acute, as unchecked bias may serve to widen the digital divide, concentrating the predicted trillion-pound economic benefits of AI predominantly in the already developed world, while marginalising emerging economies. The controversy signals that the race for AI dominance must be balanced with the imperative for ethical and equitable development.
