In this video, Prompt Engineering reviews the latest update to Google’s Bard, which now uses Gemini Pro and has surpassed GPT-4 in the Chatbot Arena leaderboard. The Chatbot Arena leaderboard is a crowdsourced platform for evaluating large language models (LLMs) based on over 200,000 human preference votes. Bard with Gemini Pro now holds the second position, a significant improvement from its previous eighth position. The video explains how the Chatbot Arena leaderboard works and explores different versions of Gemini Pro. One key feature of the new Gemini Pro is its access to the internet via the API, allowing it to provide more current and detailed responses compared to GPT-4, which relies solely on its training data. The host demonstrates this by comparing responses from both models on recent events. The video concludes with a side-by-side comparison of Bard Gemini Pro and GPT-4 Turbo, highlighting the differences in their responses and personality. The host notes that the new version of Bard is not yet integrated into the Bard web interface or the Gemini Pro API available through Google AI Studio, but it can be tested on the Chatbot Arena.

Prompt Engineering
Not Applicable
July 7, 2024
Chatbot Arena
PT9M45S