It’s been a week since it was rebranded to “Gemini”. Google‘s artificial intelligence tool has received a new update called ‘.Gemini 1.5”
“Gemini 1.5 delivers dramatically enhanced performance,” Google DeepMind CEO Demis Hassabis said in a company blog post on Thursday (February 15). “This represents a step change in our approach and is based on research and engineering innovation across nearly every part of our foundational model development and infrastructure.”
Google is releasing a model known as Gemini 1.5 Pro for initial testing, which includes “breakthrough experimental capabilities in long-context understanding,” Hassabis said in the post. Ta.
The “context window” of an AI model consists of tokens, which are the building blocks used to process information. The larger the content window, the more information the model can capture and process.
Gemini 1.5 can process up to 1 million tokens, which means it can process “1 hour of video, 11 hours of audio, and codebases of over 30,000 lines of code or 700,000 words,” the post says. .
Google said the model can process more information than AI models from OpenAI and Anthropic, Bloomberg reported on Thursday.
The tool can also demonstrate “advanced understanding and reasoning tasks across a variety of modalities, including video,” the blog post said.
“For example, given a 44-minute silent Buster Keaton movie, the model can accurately analyze various plot points and events and make inferences about small details in the movie that are often missed.” Hassabis he writes.
Gemini was known as “bard” before the rebrand announced last week. Along with the name change, Gemini Advanced and its new Google One AI premium plan debuted, and the Gemini and Gemini Advanced mobile experience is now available as part of the new Gemini app on Android and Google apps on iOS.
Earlier this week, the company acknowledged the shortcomings of its generative AI system and pledged to address them.
One big concern is hallucinations, where AI systems confidently generate false statements.
Eli CollinsVP of Product Management at Google DeepMind proposes a solution that allows users to verify information and determine its trustworthiness by easily determining the source of information provided by AI systems. did.
For all PYMNTS AI and digital transformation coverage, subscribe to our daily newsletter. AI and Digital Transformation Newsletter.
