AI

Google’s AI “Bard” is now Powered by Gemini, the most advanced AI model yet

3 Min Read

Google has released a big update for its generative AI chatbot, Bard. Now integrated with Gemini, this is the latest and most sophisticated AI model with great and enhanced capabilities. Gemini, available in Ultra, Pro, and Nano variants, is designed to operate seamlessly across a spectrum of devices, from mobile platforms to expansive data centers.

The deployment of Gemini into Bard will unfold in two phases. First, Bard will receive an upgrade featuring a finely tuned version of Gemini Pro. Next, Google will later on down the line, introduce Bard Advanced next year, which will grant users access to the pinnacle AI model, starting with Gemini Ultra.

Before its public launch, Gemini Pro underwent rigorous testing against industry benchmarks, with notable success. In six out of eight benchmarks, Gemini outperformed GPT-3.5, showcasing superior performance in tasks such as massive multitask language understanding (MMLU) and grade school math reasoning (GSM8K).

While this is an impressive performance, we’ve got to understand that GPT-3.5 is over a year old. So technically, Gemini Pro is more of a catch-up move than an outright outperformance. But if we match it with GPT-4, then we are talking.

Besides OpenAI plans to release a more powerful version of its software next year, with CEO Sam Altman expressing confidence that the new products will surpass the current ones significantly.

However, the enhancements brought by Gemini promise to make Bard more proficient in understanding and summarizing content, reasoning, brainstorming, writing, and planning, according to Sissie Hsiao, VP and GM of Assistant and Bard at Google.

Gemini Pro will first power text-based prompts in Bard, with plans to expand to multimodal support, including images, in the coming months. The year 2024 will witness the debut of Bard Advanced, featuring Gemini Ultra, the most capable model, equipped with multimodal reasoning capabilities.

Gemini Ultra can comprehend and act upon diverse types of information, including text, images, audio, video, and code, while also demonstrating proficiency in understanding, explaining, and generating high-quality code in popular programming languages.

Google aims to launch a trusted tester program for Bard Advanced before a broader release early next year, accompanied by additional safety checks. This update follows a series of improvements to Bard within its eight months of existence, including features such as answering questions about YouTube videos, integration with Google apps, and enhanced fact-checking capabilities.

Describing this update as the most significant quality improvement since Bard’s launch, Hsiao highlighted the company’s commitment to advancing AI as Bard is evolving toward its full potential. For a start, Bard with Gemini Pro will only be accessible in English across more than 170 countries and territories worldwide. More language expansions, including the European Union and the United Kingdom, are on the way.

Leave a comment

Leave a Reply

Your email address will not be published. Required fields are marked *