Google unveils three new “Open” generative AI models | Generative ai Benefits for Business Examples | Generative ai Financial Services | Microsoft Generative ai Tools | Turtles AI

 Google launches new generative AI models focusing on security and transparency

 Key points:
- Gemma 2 2B: Lightweight model for analytics text generation.
- ShieldGemma: Security classifiers to filter inappropriate content.
- Gemma Scope: Tool to make the inner workings of Gemma 2 models interpretable.
- Accessibility: Models available on platforms such as Vertex AI, Kaggle and AI Studio.

Google has announced the release of three new generative AI models belonging to the Gemma 2 series, designed to be more secure, more compact and more transparent than most available alternatives. The new models, named Gemma 2 2B, ShieldGemma and Gemma Scope, are intended for different applications, sharing a common commitment to security.

 Gemma 2 2B: Power at your fingertips

The Gemma 2 2B is a lightweight analysis text generation model designed to run on a wide range of hardware, from laptops to edge devices. This model is licensed for specific commercial and research applications, making it accessible through Google’s Vertex AI model library, the Kaggle data science platform, and Google’s AI Studio toolkit. Its light weight does not compromise power, offering a versatile solution for developers and researchers.

 ShieldGemma: Security First

ShieldGemma is a collection of security classifiers, based on Gemma 2, designed to detect and filter toxic content such as hate speech, harassment and sexually explicit materials. This tool can be used both to filter prompts in a generative model and to analyze generated content, providing an additional layer of protection for end users.

 Gemma Scope: Clarity in internal processes.

The Gemma Scope model is intended to make the inner workings of Gemma 2 models more interpretable. Using specialized neural networks, Gemma Scope decompresses the dense information processed by Gemma 2, making it easier to analyze and understand. This tool allows researchers to gain valuable insights into how the model identifies patterns, processes information and makes predictions, contributing to greater transparency and understanding of the inner workings of AI.

 A step toward transparency

Google’s Gemma model series represents a significant effort to promote transparency and safety in the use of AI. Unlike the Gemini series, whose source code is not available, the Gemma models are designed to be accessible and usable by the developer community, similar to what Meta is doing with its Llama models.

The release of these new models follows the U.S. Department of Commerce’s recent approval of open AI models, highlighting the importance of expanding access to generative AI for smaller companies, researchers, nonprofits and individual developers. However, the report also emphasizes the need for adequate capabilities to monitor these models and manage the potential risks associated with them.

 Availability and the future

The new Gemma 2 models are already available on several platforms, offering new opportunities for developers and researchers to explore and use generative AI in a safe and transparent way. With this move, Google continues to promote responsible and accessible use of AI, opening new avenues for innovation and collaboration in the global technology community.