Today’s Cache | Google’s Gemini image gaffe hangs over AI sector
| Photo Credit: REUTERS

(This article is part of Today’s Cache, The Hindu’s newsletter on emerging themes at the intersection of technology, innovation and policy. To get it in your inbox, subscribe here.)

Google’s Gemini image gaffe hangs over AI sector

After Google’s Gemini generated images of racially diverse Nazis and the company stopped the model from creating images of people, the shadow of the gaffe hangs over the AI sector as Google officials and technologists try to understand what went wrong – and hope to stop it from happening again. Google co-founder Sergey Brin acknowledged at an AI hackathon that the company had “definitely messed up” with regards to image generation and that it should have tested Gemini more rigorously.

Other entrepreneurs and executives felt that whether the reactions were overdone or not, the widely reported errors pointed to the problem of a few tech giants dominating the AI space as companies race to release new products and services. Google’s Gemini mistakes also highlighted the need to accurately take AI’s cultural biases into account without either ignoring or over-compensating for them.

ByteDance caught in the U.S.-China tech cold war

A little more than 10 years ago, Zhang Yiming, former roommate Liang Rubo, and a group of developers started TikTok-parent ByteDance, which would evolve to become a global behemoth and occupy a place of importance in the U.S.-China tech cold war. Now, American lawmakers want ByteDance to divest its U.S. stake in TikTok, or they plan to see the app banned. The U.S. is one of TikTok’s top markets and its popularity triggered other social media rivals to focus on short video offerings.

The rise of ByteDance and TikTok can be largely credited to an advanced AI-based recommendation engine, making lawmakers and technologists worry about addictive feeds. There were also suspicions regarding the degree of control China has over ByteDance, and whether the Chinese government could access Americans’ data for the purpose of espionage.

SpaceX working on spy satellite network

Elon Musk’s SpaceX is working on a network consisting of hundreds of spy satellites for the U.S. government, according to anonymous sources. The contract is reportedly classified, worth $1.8 billion, and was signed in 2021 with the National Reconnaissance Office (NRO). Musk made SpaceX’s satellite-based internet Starlink available to Ukraine’s government as its military was fighting off a Russian invasion in 2022, and it appears as though the U.S. Pentagon is also interested in SpaceX’s low-Earth orbiting satellite systems that could support military use cases.

Official statements are yet to come from SpaceX, while the NRO confirmed that it was developing a world class “space-based intelligence, surveillance, and reconnaissance system.”

Source link