Bergur Thormundsson
Research expert covering technology in the Nordics, AI and smart cities
Gemini - statistics & facts
Gemini leaped into the ring of the generative artificial intelligence (AI) market in early 2024 following the rebranding of Bard. Bulked up and trained by Google, one of the world’s largest tech companies and a giant in the AI field, Gemini has poised itself ready to take on any challengers. Indeed, early benchmarking shows the generative AI tool being capable of outdoing its toughest opponents in reasoning tasks, the likes of ChatGPT for instance. Gemini will be made available in two forms, one mobile and the other browser-based, with capabilities depending on the mode of usage.
In May 2024 Google announced that the AI program would become available as an assistant to Google Workplace users, enabled in programs such as Gmail, Calendar, Google Docs, and others. There it will perform an assistive role in writing and organizing. Additionally, Gemini will also become a stable assistant to Google’s search engine, enabling users to create itineraries and give faster answers with AI assistance for easy clarification.
Swift and transformative history
Gemini Ultra is the evolved form of Google’s first generative AI tool Bard, which had been released in March 2023. Bard's initial release faced several issues. The early showcases either had flawed answers or were heavily doctored by Google staff during release. Nonetheless, Bard was still at the forefront of the generative models of early 2023, but as the market has matured, it needed a facelift – enter Gemini. The newest Ultra iteration of Gemini will be capable of all the major aspects of generative AI models, powered by deep learning technologies. It will be able to code, answer math problems, help with writing, and generate images.In May 2024 Google announced that the AI program would become available as an assistant to Google Workplace users, enabled in programs such as Gmail, Calendar, Google Docs, and others. There it will perform an assistive role in writing and organizing. Additionally, Gemini will also become a stable assistant to Google’s search engine, enabling users to create itineraries and give faster answers with AI assistance for easy clarification.
A capable and advanced model
Gemini Ultra is already measuring well in benchmarks performed on generative AI models, to the degree that it can measure up to the other top-of-the-line large language model, GPT-4. GPT-4 already outperforms other LLMs in the market, outscoring Grok-1 from xAI, as well as Anthropic’s Claude 2, among others. Gemini Ultra is performing better across all benchmarks compared with GPT-4, thus pushing it to the very top of the generative AI category. This includes tough categories such as art & design, health & medicine, and engineering, where it performed better than GPT-4 in nearly all categories.Gemini takes Google back towards the top
The performance of Gemini Ultra is likely to take Google back towards the top of the AI developers list. The company has faced stiff competition from OpenAI, funded by the enormous research budget of tech giant Microsoft, with the GPT-4 model. Regardless, the current outlook on Gemini’s capabilities means that Google, which relies on the capabilities of its AI models for its business model, is likely to have regained an even greater standing in the ongoing AI race.
