The Unexpected Perils of Wokeness

WHEN I TYPED my own name to ask Google’s new artificial intelligence (AI) assistant Gemini for a biography, the response was flattering. Among other achievements, I was in­formed to be the writer of three books that I had never heard of. I pointed out the error and it apologised and then wrote the exact same thing again. Not many will be angry when such disin­formation makes you more than what you are. Except that, going by news reports, I was in a minority of one. Ever since Gemini rolled out, it has been a public relations disaster for Google.

In India itself, it has managed to antagonise the government to the point that legal action is being threat­ened. This was because someone asked whether Prime Minister Narendra Modi was a fascist and Gemini gave a vacillatory reply listing out what the accusers claim. When the same question was asked about former US President Donald Trump, it refused to reply citing in a roundabout way the US going to elections. But India, too, has an election. Rajeev Chandrasekhar, the minister of state for electronics and technology, posted on X that what Gemini came up with violated the IT Act and criminal code. He doubled down the next day writing that excuses like “sorry unreliable” would still not exempt Google from the law. A few days later, Union IT Minister Ashwini Vaishnaw said at an event that it was Google’s responsibility to train their AI models properly and biases would not be tolerated.

In the US, Gemini has received another form of backlash, for going overboard with political correctness to the extent that it falsified history. When it was asked for an image of the founding fathers of America, who were all whites, it threw up a native American, an Afri­can American, and two other men with brown skin. For the Pope, it came with a woman. For German Nazi soldiers, it came with a Black man. The backlash led to Google suspending Gemini’s ability to generate images. A senior vice president of the company wrote a blog explaining that they wanted to correct for racial bias existing online on which the model had been trained, and that was why such responses had crept in. Its CEO Sundar Pichai wrote in an internal memo that the errors were unacceptable and that they would continue to refine the model to make it more accurate.

It still does not address the root cause of the issue—that the reason for the AI to be wrong is because of the woke mental­ity that is part of the workforce creating the product. It is what leads to the idea that promoting diversity is a big enough cornerstone for technology to bend to adapt to it. But when it does, it shows up jargon like diversity for what they are—not objective quantifiable facts but a political ideology that is fashion­able at the moment with some sections of society, among which happen to be programmers of Silicon Valley. In some ways, such episodes are useful because they reveal how anything—even pro­gressive liberal ideas—can get hollowed out by extremism.

artificial intelligenceGeminiGoogleNarendra ModiSundar Pichai