Google DeepMind’s new Gemini model looks amazing—but could signal peak AI hype
MIT Technology Review
|
Summary
Google's DeepMind unveils Gemini, competitive AI against OpenAI's GPT-4. Though surpassing GPT-4 in most metrics, the advancements marginally small. Gemini, a multi-modal AI can process various inputs (text, images, audio), and a revamped Bard chatbot incorporates it. Comes in three versions tailored for different computing resources; full release pending for comprehensive safety checks. Despite superior performance on benchmarks, Gemini falls short on image and video tasks. Increased factual accuracy and attribution usage noted, however general challenges with generative AI continue. Despite criticisms, Google's CEO, Sundar Pichai, remains optimistic about multi-modal AI's potential.
