In a groundbreaking move, Google has introduced its latest artificial intelligence model, Gemini, describing it as “its largest and most capable” AI model to date. Sundar Pichai, CEO of Alphabet, expressed his excitement about Gemini, calling it a realization of Google’s vision for the future.
Gemini 1.0 is not a singular model but a trio, each designed for specific tasks and purposes:
- Gemini Ultra: The largest and most capable model, tailored for highly complex tasks.
- Gemini Pro: Positioned as the best model for scaling across a broad spectrum of tasks.
- Gemini Nano: Recognized as the most efficient model, specifically crafted for on-device tasks.
Google has emphasized that Gemini will be seamlessly integrated into Bard, Google’s AI chatbot, and the latest Pixel 8 Pro smartphones. Sundar Pichai highlighted the significance of Gemini, stating, “This new era of models represents one of the biggest science and engineering efforts we’ve undertaken as a company.”
Gemini is the result of extensive collaboration across Google, involving teams from Google Research. It stands out for being a multimodal model, capable of understanding and combining various types of information, including text, code, audio, image, and video.
Gemini Ultra: Pushing Boundaries
Gemini Ultra, the flagship model, has achieved a remarkable milestone by outperforming human experts in Massive Multitask Language Understanding (MMLU). This involves a comprehensive range of 57 subjects, spanning math, physics, history, law, medicine, and ethics. Google claims that Gemini Ultra can comprehend nuance and reasoning in complex subjects, marking a significant advancement in AI capabilities.
The launch of Gemini comes eight months after the introduction of Bard and one year following OpenAI’s release of ChatGPT on GPT-3.5. Google executives reportedly stated that Gemini Pro surpasses GPT-3.5, while Gemini Ultra outperforms GPT-4 in certain benchmarks, as outlined in a white paper released by Google.
Gemini in Action
Google has wasted no time experimenting with Gemini’s capabilities, integrating it into its search platform to enhance the Search Generative Experience (SGE). This integration has reportedly led to a 40% reduction in latency in English searches in the United States, accompanied by improvements in search quality.
Starting December 13, developers and enterprise customers will gain access to Gemini Pro through the Gemini API in Google AI Studio or Google Cloud Vertex AI. Android developers will also have the opportunity to leverage Gemini Nano via AICore, a new system capability available in Android 14, beginning with Pixel 8 Pro devices.
The introduction of Gemini underscores Google’s commitment to advancing AI capabilities, ushering in a new era of models with enhanced multimodal functionality and promising applications across various domains. As Gemini takes center stage, the tech community eagerly anticipates the transformative impact of this latest addition to Google’s AI repertoire.
