Google launches Gemini, its groundbreaking artificial intelligence (AI) model, asserting its superiority over OpenAI’s GPT-4. The tech giant unveiled on December 6, and comes in various optimized versions, labeled Ultra, Pro, and Nano, catering to diverse sizes and applications. Google’s CEO Sundar Pichai and Google DeepMind CEO Demis Hassabis made the announcement through a blog post.
Gemini’s Multimodal Prowess
Gemini stands out with its multimodal capabilities, surpassing GPT-4, particularly in mathematics and specialized coding. Google claims that its Ultra version achieves “state-of-the-art performance” across 30 out of 32 academic benchmarks used in large language model development. Additionally, Gemini scores an impressive 90% on a massive multitask language understanding (MMLU) test, outperforming human expert performance.
Versatility and Advanced Skills
Jeff Dean, Google’s chief scientist, highlights Gemini Ultra as the first model to achieve human-expert performance on MMLU across 57 subjects. The system seamlessly reasons across text, images, audio, and video, setting it apart from competitors. Gemini is inherently multimodal, designed from scratch, rather than grafting on vision and audio encoders later.
Advanced Programming Skills
Gemini showcases advanced programming skills, excelling in generating high-quality code using AlphaCode 2, an advanced code-generation system. It tackles complex programming problems and collaborates effectively with developers. According to AI expert Rowan Cheung, Gemini Pro outperformed GPT-3.5 in six out of eight benchmarks, making it a dominant force in the chatbot market.
Google launches Gemini: See it in Action
For those eager to experience Gemini, a fine-tuned version of Gemini Pro is now available on Google’s ChatGPT, known as Bard. This major upgrade is accessible in over 170 countries and territories, with plans for expansion into different modalities and languages. Gemini is also integrated into Google’s flagship phone, the Pixel 8 Pro, powering new features like Summarize in the Recorder app and Smart Reply in Gboard.
Future Deployments
Google plans to deploy Gemini across various products and services, including Search, Ads, and Chrome, in the coming months. The tech giant is experimenting with Gemini to enhance its web-dominant search engine, promising a generative search experience.
Conclusion
Introduced earlier this year, Gemini fulfills Google’s promise of delivering a powerful AI model, surpassing the capabilities of ChatGPT and GPT-4. Its versatile applications and advanced features position it as a significant player in the evolving landscape of artificial intelligence.
