Google Sharing Select AI to Power Chatbots

Like Meta before it, Google is allowing outside programmers to access its AI code—but not the complete powerful system. Google just released the computer code powering its online chatbot, stating that the benefits outweighed the risks of sharing it.

In a recent blog, Google explained it would release two AI language models that could help developers build online chatbots like Google’s. The two models are dubbed Gemma 2B and Gemma 7B, however, the company won’t be releasing Gemini, for free. Google is now charging for access to the most powerful version of Gemini.

There are two camps, Meta and Mistral are open sourcing their code, claiming it’s a safer approach that allows outsiders to identify problems. While Google has open sourced many AI technologies, under competitive pressure from OpenAI, it started to become more unwilling to share. That’s changing with the level of interest from developers. Google wants to provide as much safety in the products as possible but admits that bad actors might still use these technologies to cause problems.

Google is allowing people to download systems trained on enormous amounts of digital text. Analyzing that volume of data requires hundreds of specialized computer chips and tens of millions of dollars—resources few have