Google has unveiled a major upgrade to its conversational AI chatbot Bard, now powering it with the company’s own new language model called Gemini, according to a recent The Verge report. With its first rollout in 170 countries today for English users, Google hopes this will make Bard a viable rival to ChatGPT and other leading chatbots built on GPT-3.5 and GPT-4 models.
Bard is now running on Gemini Pro, described as the “Goldilocks” version of Gemini that balances speed, efficiency, and capability for chatbot applications. It sits between Gemini Ultra, the largest and most advanced version for complex reasoning, and Gemini Nano, optimized for on-device tasks. Powered by Gemini Pro, Google believes Bard can now match or outdo what ChatGPT can offer across key areas like summarization, writing assistance, question answering, and more.
Multimodal strengths for images, audio, video
One distinct advantage of Gemini is its strength for multimodal applications with images, audio, and video. Built from the ground up to seamlessly integrate across modalities, Gemini enables more interactive features like getting AI feedback on photos of design prototypes or homework problems. This could allow for more visual interaction modes in the planned “Bard Advanced” release next year based on Gemini Ultra.
CEO Sundar Pichai suggests this upgrade represents a major quality leap for Bard, with improvements in understanding intent, factual accuracy, and overall response quality. But he simultaneously described it as just the start for Gemini models powering conversational AI. The goal is to keep rapidly iterating to achieve capabilities on par with OpenAI’s GPT models that dominate today.
The launch leaves Bard better equipped to challenge rivals in key chatbot performance benchmarks. But with multimodal and advanced Gemini versions still forthcoming, Google aims higher to make Bard AI competitive through continuous modeling breakthroughs. For them, today marks meaningful progress en route to that end goal.
