Google Introduces Gemini its new AI Model

SUMMARY

Google has introduced its latest AI model called Gemini, which is the first model to outperform human experts on Massive Multitask Language Understanding (MMLU). Gemini will bring AI-powered features to the Google Pixel 8 Pro smartphone and will be integrated into Google Bard. There will be three versions of Gemini, including Ultra, Pro, and Nano, each designed for different tasks. Gemini Ultra has exceeded current state-of-the-art results on 30 of the 32 widely-used academic benchmarks and has a score of 90.0% on MMLU. The model's reasoning capabilities enable it to think more carefully before answering difficult questions, leading to significant improvements. Google has delayed the launch of its next-generation AI model, Gemini, until early 2024 due to issues handling non-English prompts. The unreleased AI is rumored to have unrivaled text and image generation capabilities and is said to outperform OpenAI's GPT-4. The rescheduled launch events were originally scheduled to take place next week in New York, Washington, and California. Google representatives have not commented on the delay. Google has postponed the launch of its Gemini AI, reportedly due to inconsistent performance with non-English queries. Despite the delay, Google is allegedly preparing for a public preview of Gemini, with four models - gemini-pro, gemini-pro-vision, gemini-ultra, and gemini-ultra-vision - expected to be added to the Vertex AI Model Garden. The delay highlights the intense competition in the AI sector, particularly against OpenAI, Microsoft, and Meta. Gemini's underperformance in multilingual tasks compared to OpenAI's GPT-4 has implications for Google's AI strategy, potentially delaying updates to other products like Bard, Assistant, and Docs. Despite the setback, Google remains committed to advancing Gemini, with new models appearing in the Vertex AI Model Garden, signaling that it will launch soon. Google has reportedly postponed the launch of its Gemini AI to 2024, canceling the scheduled launch events for next week. The company initially planned to debut Gemini before the end of the year, but sources have indicated that it was struggling with non-English queries, prompting CEO Sundar Pichai to delay its release. Gemini is intended to have impressive multimodal capabilities and handle a broad range of applications, combining different types of data like images and text for more advanced tasks. It is expected to bring improvements to Google's existing AI and AI-enhanced products, but the company seems to be taking a cautious approach to ensure it is fine-tuned and rigorously tested for safety before its release. This decision may be influenced by the dominance rival OpenAI already has in the space. Google has introduced its latest AI model called Gemini, which is the first model to outperform human experts on Massive Multitask Language Understanding (MMLU). Gemini will bring AI-powered features to the Google Pixel 8 Pro smartphone and will be integrated into Google Bard. There will be three versions of Gemini, including Ultra, Pro, and Nano, each designed for different tasks. Gemini Ultra has exceeded current state-of-the-art results on 30 of the 32 widely-used academic benchmarks and has a score of 90.0% on MMLU. The model's reasoning capabilities enable it to think more carefully before answering difficult questions, leading to significant improvements. Google's new Gemini AI model has received mixed reviews after its debut, with a video demo showcasing its capabilities going viral. However, it has been revealed that the demo was faked, with carefully tuned text prompts and still images used to misrepresent the model's actual capabilities. While Gemini does appear to generate the responses shown in the video, the interactions depicted did not actually occur in real time. This has raised questions about the speed, accuracy, and fundamental mode of interaction with the model. Google's attempt to inspire developers with the video has instead led to skepticism and a lack of trust in the company's claims about its AI model.

REFERENCES

• What is Google Gemini? Google’s New AI Model Explained(original source)
• Google Introduces Gemini its new AI Model(original source)
• Google’s best Gemini demo was faked(original source)
• Google has quietly pushed back the launch of next-gen AI model Gemini until next year, report says(original source)
• Google is reportedly pushing the launch of its Gemini AI to 2024(original source)
• 4 Google Gemini Models Coming To The Vertex AI Model Garden(original source)