The Shark Tank

AI is rapidly developing, with new AI technology being launched every month. However, with this new technology comes concerns about user responsibility and harmful AI practices. Ideally, AI developers aspire to make AI accessible to the public under safe conditions. Google has introduced Gemma, an open-language AI model that can generate content through tools and chatbots from a laptop. The same technology used to develop Gemini was leveraged to create a new, effective, yet convenient, open language model.

A Look Into Gemma’s Capabilities

Two current open-source language models that have built a strong reputation for high performance are Llama 2 and Mistral 7B. These language models have massive multitask language understanding, multi-step reasoning, and Python code creation capabilities. However, Google’s research shows that Gemma outperforms their models in nearly every category, especially coding and mathematics.

Gemma can do anything these language models can, and both create content-generation tools and chatbots. Gemma has two versions, with differences in their designated parameters. The smaller version has two billion parameters, while the larger version has seven billion parameters. The seven billion parameter model can understand more complex language and generate more refined responses than the smaller model but also requires more system support.

Who Can Use Gemma?

Google’s motive for releasing Gemma was to increase AI accessibility to the public, but Google also wants to ensure that their technology is used responsibly. Under an open license, anyone can use Gemma for either commercial or non-commercial purposes. An open license permit ensures users can use the technology freely, except for harmful usage or manipulation.

Additional safety measures Google has taken include:

• Refining the training data to remove privacy-sensitive information.
• Implementing reinforcement learning techniques.
• Auditing for dangerous capabilities.

One important tool Google released to help users navigate safe AI practices is a toolkit, including a guide for best practices, safety, and a model debugging tool.

Looking at the Bigger Picture: Gemma’s Transformative Implications

Although transformer-based models on the market are open to the public, such as GPT-3 and BERT, tech giants like Meta, Google, and Microsoft have proprietary rights over their most complex models like Gemini and LaMDA. Google’s open-source launch of Gemma pressures other tech giants to follow suit and share advanced technology with the public. Whether this was Google’s intention, Google will become a model for AI inclusivity and responsibility.

Companies must now determine which technologies should be available to the public and which to commercialize. Nevertheless, big tech giants will no longer be able to dismiss societal concerns based on proprietary rights as society begins demanding accountability.

The launch of Gemma introduces a shift towards more accessibility to AI technology and brings up questions on how to establish ethical practices. As AI continues to advance and evolve, Gemma could be a constructive prototype to analyze and build off on to develop new open-source technology that is both innovative and ethical.

Reach out to BluShark Digital for more industry news as well as individualized digital marketing strategies to help you grow your online presence as well as your business overall.