As AI continues to evolve, more people are finding uses for large language models (LLMs), which can help with everything from text to image generation. However, not everyone has started leveraging AI in their daily lives. In some cases,AI and LLMsare still viewed as intimidating. Now, Mozilla wants to help knock down some of the barriers surrounding AI.
Mozilla has announcedthat it will be launching a new open source initiative called Llamafile. It explains that the goal behind it is to help people run LLMs on the hardware they already have — no heavy-duty AI-specific CPU required. Llamafile works by condensing all of the complexity of an AI chatbot down to a single file. This file can then run on six different operating systems — Windows, macOS, Linux, OpenBSD, FreeBSD, and NetBSD — and it requires virtually no installation or configuration.
With all of your LLMs condensed into one file, you’re able to easily run it on your existing computer locally, no cloud access necessary. Owning the file means that you can also transfer it onto a storage device, if you so choose, for use elsewhere. Mozilla is referring to Llamafile as local AI, explaining that you can even turn off your Wi-Fi connection and deploy it whenever you want.
While it might sound like a novel concept, Mozilla isn’t the only one thinking about how to capitalize on local LLMs. Google has notably been trying to push itself to the front of the AI evolution with its own developments. For example,the AiCore app was discoveredwithin the Pixel 8 Pro after it launched, but little was known about its functionality at the time. More details have since come out via its Play Store listing, noting that AiCore will help “AI-driven features run directly on your device using the latest foundation models.”
Google has also notably been working hard to rival developments from OpenAI, the company behind ChatGPT.Gemini is its primary competitor,which it has deemed its most capable and advanced AI model to date. The company has introduced Gemini in different scales — Ultra, Pro, and Nano — each designed to operate at a different level. For example, Gemini can be deployed on devices as small as phones or as large as data centers.
At the moment, there is no telling how much potential AI has, but there are plenty of companies looking to find out. Businesses like Google and OpenAI might have more resources to explore the potential. However, it isn’t likely to stop other entities like Mozilla from testing the waters — and the end result may be beneficial for the greater good.
