Scrutinise AI tools, but don’t throttle the tech’s growth

The apparent racial and gender biases in the responses of AI tools also raise the question of the lack of diversity in the teams that write the algorithms of these tools.
Google AI tool Gemini.
Google AI tool Gemini.

Generative AI reflects the frailties of humans, whom they are touted to surpass in intelligence and efficiency. The latest controversy over Google’s Gemini AI’s responses clearly shows the inbuilt human biases in these AI tools. Gemini has put Google in trouble in India as well as overseas. In India, its depiction of the prime minister invited the wrath of the government, which threatened legal action; Google then retracted its content. Globally, Gemini’s image generator tool has come under severe criticism for its gender and racial biases. Google had to apologise for what it described as inaccuracies in some historical image generation depictions.

Gemini, it seems, has outraged people across races, genders and political affiliations. While it answers in affirmation to the question if Narendra Modi is a fascist, it refuses to answer if Donald Trump is a fascist. The tool has shown coloured people as Nazi soldiers. Its response to the prompt “pictures of Swedish women or American women” led to comments on social media that echoed the thought “it is embarrassingly hard to get Gemini acknowledge the existence of white people”. The latest controversy over Gemini’s results has triggered a debate over the conduct of AI and credibility of AI-generated content. The apparent racial and gender biases in the responses of AI tools also raise the question of the lack of diversity in the teams that write the algorithms of these tools.

Though these are early days and AI technology usually improves in exponential leaps, the problems with generative AI could also become more sophisticated. The latest controversy strengthens the case of instituting regulations for the industry in general and generative AI in particular. But knee-jerk reactions to these early setbacks could thwart future innovations. The technology industry has largely resisted the idea of a government-appointed regulator, but they are comfortable with a self-regulatory body overseeing issues like data governance. With more good data, responses from generative AI tools are expected to get better and more accurate. For now, sourcing and copyright of the data thus sourced remains a problem for the industry. Generative AI is likely pose a bigger challenge for the authorities in the near future, but it would be ill-conceived to bulldoze a potentially revolutionary idea with a tough regulatory regime just because someone made a mistake.

Related Stories

No stories found.

X
The New Indian Express
www.newindianexpress.com