Google is gearing up to release its much-anticipated conversational AI system Gemini, setting the stage for an all-out AI battle royale with OpenAI’s ChatGPT.
According to a report by The Information, Google has opened up testing of Gemini to select companies, indicating its consumer launch could be imminent. Gemini is a multimodal LLM, which means that it can natively receive inputs of different modalities —for example, text, image, audio, etc.
“Gemini was created from the ground up to be multimodal,” Google CEO Sundar Pichai said at the Google i/o keynote of in May. While still early, we’re already seeing impressive multimodal capabilities not seen in prior models,” he added.
This would pit the tech titan against OpenAI’s smash hit chatbot ChatGPT, which has been making serious waves since its launch last November. From fintech to healthcare, businesses are already shelling out big bucks to tap into its uncanny conversational abilities.
But Pichai appears unruffled by the prospect of an AI duel.
“It’s not fully clear to me that it might have worked out as well,” Pichai told Wired when asked if Google should have put out a ChatGPT competitor sooner.
Pichai has been ramping up investment in what he calls Google’s “AI-first future” since 2016. But he believes more time was needed to refine its models before primetime. “I feel very comfortable about where we are,” he said.
Google’s initial launch of its LaMDA-based chatbot Bard in February was a bit of a flop after it flubbed basic facts about the James Webb telescope. But Bard has since gotten a major upgrade to tap into Google’s new multimodal model PaLM 2, which “significantly outperforms” its predecessor, as Decrypt could confirm in a comparison.
The revamped Bard has already proven it can go tit-for-tat with ChatGPT Plus in certain areas while being free and more widely accessible. A Decrypt test found Bard delivered faster and more contextual translations compared to ChatGPT. It also crushed the competition when generating code, supporting more than 20 programming languages. But ChatPGT has caught up in that regard with its Advanced Data Analytics tool, which has turned the chatbot into a powerful coding machine that’s now way more advanced than Bard’s native capabilities.
And Bard’s integration with Google’s vast real-time search index gave the chatbot an edge in answering current event questions. Google is also the main investor in Anthropic, the startup behind Claude AI, ChatGPT’s most powerful competitor to date.
Claude can handle over 100K tokens of context, which means users can give the chatbot 10 times more information than what ChatGPT can handle. And the results regularly place it between GPT3.5 and GPT-4 in terms of quality. With Claude’s large context capabilities, Bard’s integration with Google for the average consumer and Gemini’s multimodal capabilities, Google is looking to dominate the AI space and take the throne currently occupied to OpenAI and Microsoft.
Google is also pursuing an “ecosystem approach” to AI, integrating generative capabilities across its products like Docs, Maps and Search. Just this week it announced new AI upgrades aimed at improving how people find and digest information online.
With Gemini waiting in the wings, OpenAI’s AI empire could soon face its biggest challenge yet. But for now, it appears Pichai is happy to take a slow and steady approach.