Google has unveiled its next-generation artificial intelligence model, called Gemini 1.0. Gemini is what is known as a multimodal model, meaning it has the ability to understand and work with multiple types of data, including text, computer code, audio files, images, and video. Having multimodal understanding allows Gemini to have more sophisticated reasoning, coding, content summarization, and planning capabilities compared to AI systems trained solely on text data.
Specifically, Google says benchmarks show that versions of Gemini like Gemini Ultra outperform other leading AI language models, like GPT-4, on assessments of language proficiency, mathematical reasoning, coding skills, and multimodal testing. This means that Gemini represents a huge leap forward in AI core competencies.
As part of the launch of Gemini, Google is integrating a streamlined version called Gemini Pro into its conversational AI service called Bard. Google claims that adding Gemini Pro drives the biggest quality update to Bard since Bard was first released to the public. The integration dramatically improves Bard’s skills in areas such as logical reasoning, writing coherence, content summary, and more.
Over the course of 2023 and 2024, Google plans to incorporate different versions of Gemini into more products and services, including search, Chrome, advertising, and others. This will allow more real-world applications to benefit from Gemini’s advanced AI capabilities in the future.
Categories: Technology
Source: vtt.edu.vn