Apple Unveils Machine Learning Framework; Will it Compete in the AI Arena?

Google made a significant announcement yesterday regarding its artificial intelligence (AI) plans. The company’s DeepMind division introduced Gemini, its largest AI model, which serves as a multimodal foundation model. However, Google was not the only tech company taking a major step towards AI. Apple seems to have entered the AI competition in Big Tech by quietly launching MLX, a new machine learning (ML) framework. Several reports indicate that Apple could utilize this framework to develop its own AI foundation models, a speculation that had been circulating earlier this year.

According to a report by The Verge, MLX is “a machine learning framework that allows developers to build models that run efficiently on Apple Silicon and the deep learning model library MLX Data.” This is the biggest hint that this framework might be intended for developing our own AI models, though it’s unlikely to be a generative AI model like those released by Google, OpenAI, and others. Apple is pretty tight-lipped when it comes to the AI capabilities of its devices, referring instead to machine learning. Despite adding a few iPhone features, such as Personal Voice, which is largely based on AI algorithms, the company refrained from using these terms.

Who can use the Apple MLX tool?

According to a Computerworld report, MLX is not a consumer tool, but a tool for its developers to provide an efficient environment for training ML models. The interesting part of the report is that Apple hasn’t forced any particular coding language on this framework, giving developers the freedom to choose whatever language they want, and it apparently came up with powerful LLM tools in the process.

Awni Hannum, an ML researcher with Apple, posted on X and said, “Just in time for the holidays, today we’re releasing some new software from Apple’s machine learning research. MLX is a powerful machine learning framework designed specifically for Apple’s silicon (aka your laptop!).

The framework is shared on GitHub and is kept open source so developers can see and work with it freely. The MLX framework works with PyTorch, ArrayFire or Jax frameworks.

A note accompanying the release stated: “The patch is intended to be user-friendly, yet powerful for training and deploying models….We intend to make it easy for researchers to extend and improve MLX with the goal of quickly discovering new ideas.

We will only find out in due course what kind of foundation models will be created with this framework, which will clarify the direction of this release.