As demand grows on Google to explain how it plans to monetize artificial intelligence, the corporation is releasing what it believes to be its largest and most powerful model on Wednesday.
Three sizes will be available for the large language model Gemini: Gemini Ultra, the largest and most competent category; Gemini Pro, which scales across various jobs; and Gemini Nano, which is reserved for specialized workloads and mobile devices.
For now, the business intends to license Gemini to clients via Google Cloud so they can utilize it in their apps.
Developers and enterprise clients can use Google Cloud Vertex AI or Google AI Studio’s Gemini API to access Gemini Pro as of December 13.
Gemini Nano will also be available for Android developers to use. Additionally, Gemini will power Google products like its Bard chatbot and Search Generative Experience (SGE), which attempts to provide conversational-style text responses to search queries (though SGE is not yet generally available).
Businesses and corporations utilize it to discover product advertising trends and more sophisticated customer care engagement through chatbots and product recommendations. Gemini can also be used for productivity apps that need to produce code for developers or summarize meetings, or for content production for a business wanting to write blog posts or marketing campaigns.
The company provided instances, such as how Gemini might snap a picture of a chart, analyze hundreds of research pages, and then update the image. Another example was examining a photo of someone’s arithmetic assignment and pointing out the right answers and erroneous ones.
In a blog post on Wednesday, Google CEO Sundar Pichai highlighted Gemini as the outcome of extensive collaboration among Google teams, including those at Google Research. Pichai explained that Gemini, constructed to be multimodal, can comprehend, operate across, and integrate diverse forms of information such as text, code, audio, image, and video.
As of today, Google’s chatbot Bard will leverage Gemini Pro to enhance its advanced reasoning, planning, and understanding capabilities. Additionally, a forthcoming update named “Bard Advanced,” utilizing Gemini Ultra, is scheduled for release early next year, marking a substantial enhancement to Bard, Google’s ChatGPT-like chatbot.
This development follows the launch of Bard eight months ago and ChatGPT by OpenAI a year prior. Despite outperforming GPT-3.5, Google executives refrained from directly comparing Gemini Pro to GPT-4. However, Gemini’s Ultra model demonstrated superior performance in certain benchmarks compared to GPT-4, as detailed in a white paper released by Google.
Regarding potential charges for access to “Bard Advanced,” Sissie Hsiao, Google’s general manager for Bard, mentioned focusing on delivering a positive user experience and clarified that monetization details are not finalized.
During a press briefing, Eli Collins, Google DeepMind’s vice president of product, acknowledged the likelihood of novel capabilities in Gemini compared to current-generation large language models (LLMs). Collins stated that they are still exploring the unique features of Gemini Ultra.
Gemini’s launch was reportedly delayed due to readiness concerns, resembling Google’s earlier AI tool rollout challenges this year. Collins explained that testing advanced models requires more time, emphasizing Gemini as the most rigorously tested AI model with comprehensive safety evaluations at Google.
Despite being the largest model, Gemini Ultra is stated to be significantly more cost-effective to serve, offering both increased capability and efficiency. Collins highlighted the ongoing efforts to enhance computational efficiency in training such models.
In addition to the Gemini announcement, Google introduced its next-generation tensor processing unit, TPU v5p chip, for training AI models. The chip is claimed to deliver improved performance for the price compared to the TPU v4 announced in 2021.
Google’s move comes amid competition with cloud rivals Amazon and Microsoft, unveiling custom silicon targeting AI. Investors have pressed Google for details on converting AI efforts into profit during the third-quarter earnings conference in October.
In August, Google launched the “Search Generative Experience” (SGE) as an early experiment, providing a more conversational AI experience within the search engine. Despite inquiries about SGE’s timeline, Google executives were vague about its public launch, stating that Gemini would be incorporated into it “in the next year.”
Pichai expressed enthusiasm for Gemini’s potential in the blog post, characterizing it as a significant science and engineering undertaking for the company, unlocking new opportunities for people globally.