DeepMind's Gemini Aims to Rival OpenAI's ChatGPT with AlphaGo Techniques

OpenAI's ChatGPT is facing potential competition as DeepMind, the renowned research lab owned by Google, has announced that it is working on a next-gen chatbot called Gemini. By integrating techniques from AlphaGo, the AI system that made history by defeating a professional human Go player, Gemini aims to challenge and potentially surpass ChatGPT's capabilities.

DeepMind CEO Demis Hassabis shared with Wired's Will Knight that their objective for Gemini includes abilities to plan and solve problems, in addition to analyzing text. The chatbot will employ the expertise of AlphaGo-type systems, coupled with groundbreaking language capabilities exhibited by large models, to achieve this goal. Hassabis also hinted at new innovations that will contribute significantly to Gemini's capacities.

It is speculated that Gemini, which was briefly featured at Google's I/O developer conference in May, will utilize advancements in reinforcement learning to tackle challenges currently faced by existing language models. Reinforcement learning focuses on rewarding AI systems for specific behaviors while penalizing undesired ones. This approach aims to guide systems towards exhibiting suitable behaviors in various situations.

Reinforcement learning has contributed to progress in language model technology, playing a vital role in how systems such as ChatGPT respond to prompts. DeepMind, having extensive experience in reinforcement learning, is undoubtedly keen on applying its insights to the generative AI sector.

While Gemini would not be DeepMind's initial attempt at language models, it remains the most ambitious project to date. Previously, the lab introduced Sparrow, a chatbot designed to provide safer and more appropriate responses than other models. Hassabis had mentioned plans for a private beta release of Sparrow in a January interview with Time, although it is unclear if this plan has proceeded as intended.

The Information reported in March that Google executives, including its senior AI research executive Jeff Dean, are taking an active role in Gemini's development. Initially prompted by the shortcomings of Bard, Google's chatbot project, in keeping up with ChatGPT, Gemini represents a vigorous effort to compete in the generative AI market. The rapidly growing market for generative AI, which includes text-analyzing AI like Gemini, is estimated to be valued at $109.37 billion by 2030, marking a 35.6% increase from 2030 (source: Grand View Research).

Platforms such as AppMaster can help businesses and developers streamline their app development processes using no-code and low-code solutions. With AppMaster's innovative approach, technical debt is eliminated as requirements change, empowering users to create scalable software solutions complete with backends, web interfaces, and native mobile apps. Check out AppMaster's guide for insights on how to create an app efficiently.