In its first 24 hours, Google’s new generative AI model Gemini has generated a significant buzz among users. Gemini, described as Google’s most capable and general model yet, has been eagerly anticipated since OpenAI’s release of ChatGPT last year. With three different versions – Ultra, Pro, and Nano – Gemini is a multimodal AI system that processes text, code, audio, images, and video to provide responses to user prompts. Its integration into mobile devices, a first for an AI model, has also garnered excitement from app makers.
Gemini’s performance has been impressive so far. Google claims that the Ultra model, set to launch next year, exceeds current state-of-the-art results on 30 out of 32 widely-used academic benchmarks for large language models. Although the outperformance against GPT-4, the latest language model behind ChatGPT, is only marginal, initial reactions suggest that users are impressed with Gemini’s capabilities.
One user, Mihir Patel, compared responses from Gemini and GPT-4 to a deep learning question and found that Gemini provided a more detailed response, including links to external research papers. Patel exclaimed, Gemini is so good. SO much better and SO much faster than GPT-4. Another demonstration showcased Gemini accurately describing a developing picture, leading to playful responses on social media about Google’s comeback.
Gemini also presents itself as an appealing alternative to OpenAI’s offering for developers. As the Nano model is designed to run natively on Google Pixel phones, Android developers can more easily build AI apps.
While the success of Gemini remains to be seen and whether it can lure users away from ChatGPT is uncertain, the initial feedback points to a positive reception. However, Gemini still faces challenges with hallucination issues similar to those encountered by ChatGPT.
Google’s CEO Sundar Pichai, Google DeepMind boss Demis Hassabis, and other executives are aware of the innovator’s dilemma and the need for ongoing product development to maintain market leadership. The positive early reactions to Gemini suggest that Google’s search engine still has much to offer.
Although the verdict on Gemini’s long-term success is still pending, it is undoubtedly an exciting development in the world of AI. As Google continues to work on non-English language versions of the models, users eager to test Gemini may have to exercise patience. Nevertheless, Gemini’s potential impact on various industries and its integration into mobile devices make it a model to watch.