On Wednesday, Google unveiled Gemini, its “largest and most capable AI model”, as the tech giant stated in the announcement. Potentially it will become OpenAI’s ChatGPT’s only competitor on common metrics. The new technology will change Google’s consumer experience, from the apps to smartphones. It will be integrated into selected Google products, including the Pixel 8 Pro smartphone, and services like Bard chatbot.
“The Gemini Era–as it is called this new period as of yesterday–represents one of the biggest science and engineering efforts we’ve undertaken as a company”, said Google CEO Sundar Pinchai in a note shared to introduce the new technology.
Compared to the traditional and existing AI models, Gemini can handle a broader spectrum of inputs seamlessly integrating text, images, audio, video, and even programming code into its processing capabilities. It improves audio and video capabilities of Bard chatbot. Later, it will be added to Google’s search engine and Chrome browser in order to make images and sounds faster to be found and analyzed.
Gemini can mimic human reasoning and reportedly outperform experts, and most importantly for the company ChatGPT, in subjects like math, physics, and law. Google employees demonstrated it on the launching day: they used it to scrutinize math assignments and, allegedly, delved into hundreds of research papers to refine a chart. Moreover, they implemented Gemini with a new generation of powerful cloud-based processors, since cloud computing is a critical resource for AI developers.
The company is more than aware of the persistent risk of AI models misleading users. Google developers said that the company has done “a lot of work on improving factuality in Gemini”, but large language models “are still capable of hallucinating”, which means that AI systems sometimes make up facts and get information wrong, but with extreme confidence to answer to user’s requests. So this only intensifies the concerns by policymakers and AI sceptics.
