AI Wars May Have Ceasefire Agreement Sooner Than Expected

A few months ago, everyone wondered who would win the AI ​​arms race. Microsoft has associated itself with OpenAI. Google launches Bard. Meta began work on its own larger language model, LLaMA. Other companies started thinking about launching AI platforms, and curious users pitted the models against each other.

But a recent deal shows that we may see more than just one-on-one competition, but a growing number of partnerships. Earlier this week, Meta offered its LLAMA large language model for free under an open license and brought it to Microsoft’s Azure platform. The decision highlights the benefits of interoperability in AI – and as more companies join the field, it likely won’t be the last of its kind.

To date the well-known LLM has been relatively hushed up and offered in a more controlled environment where users are required to have permission to build with the model or use the data. OpenAI continues to train GPT and released GPT-4 in March Providing paid API access to developers For the latest version of its model. Apple is developing its own LLM, called Ajax, though details are scarce; It is not yet publicly available, and its open-source status is unknown. Bard, Google’s LLM, is not open source at all.

The LLaMA was initially not publicly available and only accessible through META, and META has not yet disclosed its training data. But LLAMA was always intended to be open source and was created to “more democratize access” to AI. This week, Meta at least partially fulfilled its promise. Users of the closed system must pay a license fee to access that model and distribute applications using the same model. That inconvenience went away because of the way Meta made LLAMA available to Azure users and somewhat unlicensed.

Meta open-sourcing LLaMa and bringing it to Azure makes business sense, especially if Meta openly believes in developing AI. This is the first step towards giving people access to more LLM models on the platform and the convenience of comparing results. A large variety of LLM frameworks to choose from also focus on the question of how each model can work together. And LLM developers want people to use their models, so making them available on different platforms allows them to reach more users.

Even the most competitive Big Tech companies do business with each other. Working with Microsoft is nothing new for Meta — Meta brought Microsoft’s Teams product to Workplace by Meta, which already runs the Office 365 suite.

Openness has its risks. explained Ilya Sutskever, co-founder and chief scientist of OpenAI, which was a more open organization when it was founded in 2015. ledge He regrets sharing the research over competition and fear of security. For example, opening up datasets makes it easier to sue for copyright infringement, because people can see what sources were scraped for data to train models.

But having more LLM frameworks to choose from could be good news for proponents of AI interoperability.

Since LLMs are, by default, different from each other, developers often have to choose which model to build an app with. There’s no nice way to talk to the system.

Walled gardens are no shock to most modern tech users, but AI interoperability proponents argue that the only way AI can grow and develop is not through closed silos but through open structures that can talk to each other. Even Microsoft believes in interoperable AI; It joined other tech companies in joining open neural network exchangeA group that wants to promote an industry standard for AI interoperability so that developers can “find the right combination of tools.”

AI systems working together can produce better results for things like search queries. Companies that can train models on different datasets can provide better, more complete service – and, if one model is wrong, avoid potentially disastrous overreliance on a single source of information. And being able to develop GPT models from both LLaMa and OpenAI in one place can cut development costs and timelines.

For now, having LLaMa available on Azure doesn’t mean that apps built from LLaMa can suddenly talk to those running OpenAI’s GPT model. No one has built that bridge yet. Too, not everyone agrees LLaMa checks all the boxes for open-source software, especially since it doesn’t use a license approved by the Open Source Initiative. It also limits who can use LLaMa commercially without charge. to this community license agreementDevelopers who have more than 700 million monthly active users must “request a license from Meta.”

But it’s a good step in the right direction for open source and interoperability, if only by allowing developers easier access between models. There is room for healthy competition, but if companies really want to develop AI, working together is the best option.