Google unveils 'open' AI models Gemma following Meta's lead

Google has announced the release of new artificial intelligence (AI) models, offering outside developers the opportunity to leverage and customise them for their own projects. The Alphabet subsidiary unveiled its "open models" named Gemma, providing individuals and businesses with access to essential technical data, including model weights, free of charge.

Unlike a full open-source approach, Google retains some control over the terms of use and ownership of Gemma. However, the release signifies a major step towards democratising access to AI technology, potentially attracting a wave of software engineers eager to innovate using Google's resources. The move also aims to bolster the adoption of Google's cloud division, which has recently turned profitable.

The Gemma models, optimised for Google Cloud, come in sizes ranging from two billion to seven billion parameters, defining the complexity and capabilities of the algorithms. Moreover, first-time cloud users leveraging Gemma models on Google Cloud are eligible for $300 in credits, incentivising experimentation and adoption.

Google's decision to withhold its larger, premier models, known as Gemini, from the open release sets it apart from Meta's strategy with its Llama 2 models, which range from seven to 70 billion parameters in size. While Google has not disclosed the exact scale of its largest Gemini models, it is evident that the tech giant is strategically positioning Gemma as a gateway for developers to access AI capabilities without the complexity of managing extensive model sizes.

Collaborating with chipmaker Nvidia, Google ensures seamless integration of Gemma models with Nvidia's chips, enhancing performance and scalability. Nvidia, in turn, announced plans to extend support for Gemma to its chatbot software, designed to run AI models on Windows PCs, further expanding the accessibility and applicability of Google's AI technology.

The move underscores the ongoing evolution of AI accessibility and collaboration within the tech industry. While concerns over the potential misuse of open-source AI persist, proponents argue that democratising access to such technologies is essential for fostering innovation and inclusivity.

(With Reuters inputs)