Google has revealed Gemini, its latest AI model, touting it as their “most expansive and proficient AI model” to date. This week, Google DeepMind, the company’s AI research lab, introduced the much-anticipated model, which reportedly outperforms OpenAI’s GPT-4 on significant benchmarks.
Since the launch of OpenAI’s ChatGPT a year ago, prominent tech companies have been in a race for AI progress. Microsoft had a slight advantage due to its investment in OpenAI, providing access to its models. Google had been relatively slow in this space, with the initial release of its ChatGPT competitor, Bard, facing hitches. However, with Gemini, Google aims to potentially leapfrog ahead of OpenAI.
Understanding Gemini
Gemini is being lauded for its multimodal capabilities, advanced reasoning, and coding abilities. Unlike other models built first on text and then integrating image data, Gemini is inherently multimodal. It was pre-trained on audio, image, and text modalities from the start, allowing it to comprehend and reason about various inputs better than existing multimodal models.
Scoring 90 percent in contrast to GPT-4’s 86.4 percent on the MMLU (massive multitask language understanding) benchmark, Gemini showcases its prowess in multidisciplinary knowledge and problem-solving.
In practical applications, Google has optimized Gemini for three different sizes: Ultra, Pro, and Nano. Gemini Nano is set to run on the Google Pixel 8 Pro, empowering features such as summarizing transcripts in the Recorder app and suggesting responses for Smart Reply on the phone’s keyboard (Gboard).
While Gemini Pro powers Bard for text prompts currently, Google plans to launch a revamped Bard next year, driven by Gemini Ultra. This ongoing development promises an all-around enhancement in reasoning, planning, and understanding.
Topics
Artificial Intelligence
Google
Cecily is a tech reporter at Mashable who covers AI, Apple, and emerging tech trends. Before getting her master’s degree at Columbia Journalism School, she spent several years working with startups and social impact businesses for Unreasonable Group and B Lab. Before that, she co-founded a startup consulting business for emerging entrepreneurial hubs in South America, Europe, and Asia. You can find her on Twitter at @cecily_mauran.
