Earlier this year, Google faced heavy criticism for the factual errors made by its AI chatbot, Bard. However, over time, Bard’s responses improved and the complaints about its errors decreased. Now, Google is preparing to launch its new large language model, Gemini.
Recent reports indicate that the launch of Gemini, which was expected to take place this week, has been postponed due to subpar responses to certain queries. There are also rumors that Gemini may be launched earlier than anticipated, but Google has not yet confirmed anything. Until an official announcement is made, it is best to approach this information with caution.
According to sources familiar with the matter, Google’s highly anticipated launch of Gemini has been delayed until January. The decision to postpone the public introduction was made by Google’s CEO, Sundar Pichai, who recently canceled a series of events scheduled for the upcoming week in California, New York, and Washington. The company identified reliability issues with the AI’s handling of non-English queries, prompting the need for further refinement before the launch. These events were intended to be Google’s most significant product launch of the year and were kept confidential until now.
About Gemini:
When Gemini was unveiled at Google’s I/O event in May, it was initially seen as Google’s response to OpenAI’s popular AI tool, ChatGPT. However, DeepMind CEO Demis Hassabis later stated that Gemini would surpass ChatGPT in terms of power once it is launched.
According to a report by Wired, Hassabis revealed that Gemini would be more advanced than ChatGPT and that the DeepMind team drew inspiration from their AI program, AlphaGo.
AlphaGo, developed by DeepMind, made history in 2016 by defeating a human champion in the strategic board game Go.
Hassabis explained, “In essence, Gemini combines the strengths of AlphaGo-like systems with the exceptional language capabilities of large models.” He further emphasized that Gemini would play a significant role in Google’s response to the competitive landscape, including the threat posed by ChatGPT and other generative AI technologies.
When Gemini was introduced, Google stated in a blog post, “We are actively working on Gemini, our next-generation model designed to be multimodal, highly efficient in terms of tool and API integrations, and built to facilitate future innovations such as memory and planning. Although Gemini is still undergoing training, it has already demonstrated unprecedented multimodal capabilities compared to previous models.”
The blog post also mentioned that once Gemini is fine-tuned and thoroughly tested for safety, it will be available in various sizes and capabilities, similar to PaLM 2, to ensure its widespread deployment across different products, applications, and devices for the benefit of all users.