Image: NurPhoto/Getty Images.
Google unveiled Bard, its chatbot powered by artificial intelligence, last February. With the goal of becoming a formidable competitor to ChatGPT, Bard has been powered since its release by two different major language models (LLM). And today, he benefits from the most important update to his LLM: Gemini.
What is Gemini Pro?
This Wednesday, Google released Gemini, the most powerful and advanced language model to date. It was first announced last May, at Google I/O. Along with the release of the new LLM, Google announced a specially optimized version of Gemini Pro in Google Bard for English speakers.
Gemini comes in three different sizes: Gemini Ultra, Gemini Pro and Gemini Nano – each model specializes in different tasks. For example, Gemini Nano is most effective at performing tasks on devices. Gemini Pro is the widest: it is capable of carrying out many diverse tasks. Gemini Ultra, on the other hand, is the most advanced. It is therefore best suited to complex tasks.
Thanks to Gemini Pro, Google Bard should have more advanced reasoning, planning and understanding capabilities.
Not yet available in France
To demonstrate the effectiveness of Gemini Pro, Google tested it against a number of industry standard benchmarks. Its results showed that in six of eight benchmarks, Gemini Pro outperforms GPT-3.5, the OpenAI LLM that powers the free version of ChatGPT.
Image: Google.
Google continues by indicating the result of the blind evaluations it conducted: Bard with Gemini Pro was named the favorite free chatbot of those who participated.
Bard with Gemini Pro is already available for free in 170 countries – but not yet in France – and only in English for the moment. However, Google plans to expand it to other languages and other regions of the world soon. To test it, simply use Google’s chatbot as you normally would.
Waiting for Bard Advanced
Bard won’t stop with this update. Early next year, Google plans to introduce Bard Advanced, which will be powered by the most advanced version of Gemini – Gemini Ultra. The web giant specifies that this version is capable of understanding and acting quickly on multimodal inputs – text, images, audio, video and code in particular.
Before launching Bard Advanced, Google will first conduct extensive security checks and launch a trusted tester program.
When it launched, before Gemini, Bard was powered by a stripped-down version of Google’s Language Model for Dialog Applications (LaMDA). Then LaMDA was replaced last May by PaLM 2, a more advanced version of PaLM supposed to significantly improve the chatbot experience. However, improvements so far have been rather minimal. Will the new update, then the next one planned in the program, allow Google to assert itself against OpenAI? The future will tell.
Source: ZDNet.com