The well-known ginormous high-tech search Google is moving quite fast on its model of AI (Artificial Intelligence) Gemini. Some researchers say Gemini “smashes GPT-4.” It is of paramount importance to highlight that the company is adding enterprise AI tools to its cloud platform.
SemiAnalysis, the research firm, has stated that Google has foresight Gemini AI model will crush the OpenAI offering by packing far more computing power.
Competition in AI intensifies
These claims were rejected by some experts, suggesting somehow that the data came from Google’s own internal marketing materials. One of them, OpenAI CEO Sam Altman said, “Amazing Google got that Semi-Analysis guy to publish their internal marketing/recruiting graph hahaha.” The rumors seem to indicate that Google is just creating this perception.
The semi-analysis report written by Dylan Patel and Daniel Nishball on August 28 was titled “Google Gemini Eats The World – Gemini Smashes GPT-4 By 5X.”
Said analysis has as a basis the fact that Google had access to much better processors and computing power. It has ignited a debate in AI forums about whether more computing power translates into better AI performance and all that jazz.
Researchers at Google’s Artificial Intelligence division DeepMind are working on Gemini, a next-generation, multi-modal AI model that can change many things as we know them.
Some people expect Gemini to be launched later in 2023. On the one hand, Gemini can accept text, video, audio, and images and output text and images; this is what means to be multimodal. On the other hand, ChatGPT is currently only text-based and has limitations to accessing the Internet.
In the same vein, Python programmer and AI enthusiast ‘Teknium’ commented:
“I hope someone dethrones OAI, soon. But to say Gemini Smashes GPT-4 by 5x makes it sound like it is 5x better than GPT-4, it’s not, its 5x compute. Not necessarily correlated to quality.”
While OpenAI’s ChatGPT garnered attention in 2022 and became popular, a number of different tech giants such as Google and Meta are developing powerful AI models. Thus, 2023 is shaping up to be a pivotal year for progress in LLMs.
Nevertheless, Google is already facing the music and a class-action lawsuit over the alleged misuse of consumers’ personal data used to train its AI models.
Google Cloud incorporates Llama and Claude Artificial Intelligences
In addition to competing, Google is reportedly making allegiances with companies such as Meta and Anthropic to bring their AI models to Google Cloud.
Google Cloud customers will be able to access Meta’s Llama 2 LLM, as well as AI startup Anthropic’s Claude 2 chatbot. They can customize the models with business data for their own applications and services.
Just as it has monopolized Internet search, it wants to position itself as a one-stop-shop for cloud customers to access leading generative AI.
The company is also deepening its partnership with Nvidia to accelerate computing power to develop large AI models.
By Leonardo Pérez
