Gemini language model
Google wants to outpace the competition with a new AI model
Google wants to outperform its competitor OpenAI with the new Gemini language model. photo
© Google/dpa
In the dynamic AI industry, the start-up OpenAI dominates with the text robot ChatGPT. But Google is now countering with an eagerly awaited new AI language model. What to expect?
In the race for artificial intelligence, Google wants to compete with the new one Put language model Gemini at the top. The AI system should not only keep up with the GPT4 language model of its competitor OpenAI, but also surpass it.
Google CEO Sundar Pichai said of Gemini’s announcement that the shift being experienced with AI will be the “most profound in our lifetime, far greater than the transition to mobile phones or the Internet before it.”
Gemini classifies hand gestures and drawings
Gemini can not only generate texts in a chatbot, but also solve specific problems and make situation-dependent decisions. It can also record information from photos and videos. Google demonstrated the system with a video chat in which Gemini immediately recognized and correctly classified drawings and hand gestures from the human counterpart.
“We are bringing Gemini to billions of people through Google products,” announced Google DeepMind boss Demis Hassabis. The Google subsidiary emerged in April 2023 from the British start-up DeepMind, which was acquired in 2014, and Google’s internal AI department. By integrating the start-up, which had previously operated largely independently, Pichai wanted to pool the AI efforts within the Google Group in order to be able to compete more decisively against OpenAI.
Three dimensions planned
Google will introduce the new system in three different dimensions: Gemini Ultra, Gemini Pro and Gemini Nano. Gemini Ultra is the largest and most powerful model for highly complex tasks. This variant is aimed primarily at corporate customers.
Gemini Pro will be aimed at a wide audience and, for example, teach the Google Bard chatbot “advanced thinking, planning, understanding and more”. “This is the biggest upgrade for Bard since its launch,” said Hassabis. Bard will now be available in English in more than 170 countries and territories. However, Google plans to support new languages and locations in the near future.
Gemini should also run on the Google smartphone
The third Gemini variant, Nano, brings the system to the top model of Google’s Pixel smartphones. “The Pixel 8 Pro is the first smartphone to run Gemini Nano,” announced Hassabis. For example, the Recorder app can be used to record not only spoken language from a long meeting, a lecture or an interview and convert it into written language in real time. With the help of Gemini Nano, the pixel can then create a compact summary without any time delay. In the coming months, Gemini will be available in other Google products and services such as search, ads and the Chrome browser.
Google has been working on applications based on artificial intelligence for years, but is currently under pressure to reveal more of them. A year ago, the start-up OpenAI sparked a new competition in artificial intelligence when it made its chat bot ChatGPT public. The software caused a stir because it can form sentences like a human. It is trained with huge amounts of data and estimates, word by word, how a sentence could continue. This brings with it the risk that it can give out completely incorrect information.