Google Introduces Gemma: A New Open Artificial Intelligence Model

Gemma is globally available as of Wednesday, accompanied by a new generative AI toolkit offering essential tools and guidance for developing safer applications.

To encourage adoption, Google is offering $300 in credits for first-time Google Cloud users, along with the opportunity for researchers to apply for Google Cloud credits of up to $500,000 to accelerate their projects.

Google has unveiled its latest open artificial intelligence model named Gemma, leveraging the same cutting-edge research and technology utilized in its flagship Gemini models. This initiative spearheaded by the Alphabet-owned company aims to empower developers and researchers in constructing AI applications responsibly.

Unlike Gemini, which operates as a closed AI model competing directly with Microsoft-backed OpenAI’s ChatGPT, Gemma offers a more lightweight alternative. Developed by Google DeepMind, Gemma is anticipated to excel in modest tasks such as basic chatbots or summarization jobs.

Demis Hassabis, founder and CEO of DeepMind (acquired by Google in 2014), expressed pride in the release of Gemma, emphasizing the company’s commitment to supporting responsible open-source initiatives. Gemma models, known for their exceptional performance relative to their size, draw inspiration from the technology underpinning Gemini.

Tris Warkentin, director at Google DeepMind, and Jeanine Banks, VP and GM at Google’s Developer X team, highlighted Gemma’s technical prowess and its ability to run efficiently on developer laptops or desktops. They underscored Gemma’s superiority over larger models on key benchmarks while adhering to rigorous standards for safety and responsible outputs.

Availability and Access

Gemma is globally available as of Wednesday, accompanied by a new generative AI toolkit offering essential tools and guidance for developing safer applications. Developers can access Gemma through various platforms including Kaggle, Hugging Face, Nvidia’s NeMo, and Google’s Vertex AI.

Partnerships and Collaborations

French-American company Hugging Face, renowned for its collaboration platform for the machine-learning community, has joined forces with Google Cloud to support the development of generative AI and machine learning. Hugging Face expressed enthusiasm for Google’s reinforcement of its commitment to open-source AI, pledging comprehensive integration support for Gemma.

Understanding Gemma

Launched in December, Gemini stands out as the first AI model to surpass human experts on Massive Multitask Language Understanding (MMLU), a standard method for evaluating AI’s knowledge and problem-solving abilities. Integrated with Google’s generative AI tool Bard, Gemini exhibits versatile capabilities in comprehending diverse tasks and generating code based on different inputs.

Safety Measures

Google assures that Gemma is designed with its AI principles at the forefront, incorporating automated techniques to filter out personal information and sensitive data from training sets. Rigorous evaluations, including manual red-teaming and automated adversarial testing, were conducted to mitigate risks associated with Gemma’s deployment.

Incentives for Developers

To encourage adoption, Google is offering $300 in credits for first-time Google Cloud users, along with the opportunity for researchers to apply for Google Cloud credits of up to $500,000 to accelerate their projects.

Market Outlook

Investor interest in generative AI startups has surged, with over $4.2 billion invested through 215 deals in 2021 and 2022, according to CB Insights. Globally, AI investments are projected to reach $200 billion by 2025, potentially making a significant impact on GDP, as per Goldman Sachs Economic Research. In Saudi Arabia, the generative AI market is expected to exceed $1 trillion by 2030, fuelled by a compound annual growth rate of over 25%, according to Statista data.

Industry Developments

The launch of ChatGPT by OpenAI in December 2022 thrust generative AI into the spotlight, albeit raising concerns about its usage and accuracy. In response, the Technology Innovation Institute in Abu Dhabi introduced Falcon 180B in September, aimed at advancing generative AI in the region. Amazon Web Services also entered the fray in November, unveiling a generative AI tool tailored for businesses.