Google Unveils PaLM 2 AI Model to Rival GPT-4
Google introduces PaLM 2, a powerful new large language model designed to compete directly with OpenAI's GPT-4. The model is already powering the Bard chatbot and promises enhanced reasoning, coding, and language capabilities.
Google officially unveils PaLM 2, its latest large language model that serves as a direct competitor to OpenAI's GPT-4. Announced at Google I/O, this new AI model acts as the foundational technology behind the company's expanding suite of artificial intelligence products, including the Bard chatbot.
The tech giant designs PaLM 2 to deliver expanded capabilities in reasoning, language understanding, and coding. Instead of relying solely on one massive system, Google fine-tunes the base model into specialized tools like Med-PaLM 2 for medical diagnostics and Sec-PaLM for cybersecurity threat detection.
PaLM 2 operates through four distinct submodels of varying sizes: Unicorn, Bison, Otter, and Gecko. Although Google keeps the exact parameter count under wraps, the company promises that this new architecture provides faster, more efficient performance compared to its predecessor.