Google’s New Gemini AI Just Beat Humans And GPT Across 57 Subjects

Google has unleashed its groundbreaking Gemini AI, claiming superiority over OpenAI’s GPT-4 and human experts in a stunning array of 57 subjects. Boasting a remarkable 90.0% on the MMLU test, surpassing human experts (89.8%) and GPT-4 (86.4%), Gemini emerges as a multifaceted powerhouse in math, physics, history, law, medicine, and ethics. Its prowess extends beyond text, excelling in images, video, audio, and code understanding.

Gemini’s uniqueness lies in its multimodal design, fluently navigating visual and auditory realms alongside text. Unlike predecessors, it retains the original tone and nuance when interpreting video, audio, and images. This breakthrough opens new horizons in AI, training models with diverse sensory datasets mirroring human learning processes.

The potential of Gemini transcends theoretical demonstrations, as Google plans to integrate it into devices like the upcoming Pixel phones. With the ability to comprehend touch and tactile feedback, Gemini could revolutionize AI robotics, taking interactions into uncharted territory.

One standout feature is Gemini’s proficiency in generating code, showcased in the AlphaCode 2 project. Dividing Gemini into specialized AIs for different programming phases, it competed in a coding contest, outperforming 87% of human entrants. While not immediately available to the public, this approach hints at future developments in AI-driven software teams.

Google envisions three Gemini models: Nano, fitting mobile devices; Pro, equivalent to GPT 3.5; and Ultra, surpassing GPT-4 across various benchmarks. While Ultra awaits public release, Nano is already available on the Pixel 8 Pro, and Pro can be accessed for free through Google Bard, with plans for integration into various Google products.

The promise of Gemini extends beyond conventional AI boundaries, offering a glimpse into the future of technology interactions. As Gemini integrates into Google’s product ecosystem, its impact on daily tasks, from coding to website creation, suggests a paradigm shift in human-technology interfaces. The roller coaster of AI advancement is accelerating, and Gemini appears set to redefine the landscape.