The dominance of OpenAI and Google in the machine learning and artificial intelligence space is being challenged by a rapidly growing number of open source projects that push the state of the art and leave big corporations lagging behind.

A leaked memo from within Google highlights the fact that the deep-pocketed companies have no moat and are unable to keep up with the pace of development of smaller and more nimble projects.

One of the major advantages that OpenAI and Google have over smaller players in the space is their significant funding and infrastructure. However, this head start is becoming slimmer by the day as open source projects gain ground with rapid collaboration and experimentation in subreddits and Discords.

The memo points out that a leaked foundation language model from Meta, called LLaMA, was leaked in March, and within weeks, people tinkering around on laptops and penny-a-minute servers had added core features like instruction tuning, multiple modalities, and reinforcement learning from human feedback. OpenAI and Google were probably poking around the code, too, but they didn’t — couldn’t — replicate the level of collaboration and experimentation occurring in these smaller communities.

This rapid development of smaller open source projects is challenging the business model of larger companies, like OpenAI and Google. These companies have been pursuing a SaaS model, offering access to high-value software or services through carefully gated APIs.

But customers are starting to question why they need to wade through a large language model like GPT-4, which is the largest and most general-purpose AI model ever created, when all they want to do is match the language of a contract against a couple of hundred others.

The smaller projects are proving that the titanic computation problem that seemed to pose an insurmountable obstacle to challengers is already a relic of a different era of AI development. While bigger isn’t always better, the rapid development of smaller projects shows that smaller can be just as effective, if not more so.

In conclusion, while the dominance of OpenAI and Google is not currently under threat, they will have to keep a close eye on the rapidly growing number of open source projects that are pushing the state of the art and challenging the business models of larger companies. The deep-pocketed companies must adapt and innovate in order to stay ahead of the competition.

The dominance of OpenAI and Google in the machine learning and artificial intelligence space is being challenged by a rapidly growing number of open source projects that push the state of the art and leave big corporations lagging behind.

A leaked memo from within Google highlights the fact that the deep-pocketed companies have no moat and are unable to keep up with the pace of development of smaller and more nimble projects.

One of the major advantages that OpenAI and Google have over smaller players in the space is their significant funding and infrastructure. However, this head start is becoming slimmer by the day as open source projects gain ground with rapid collaboration and experimentation in subreddits and Discords.

The memo points out that a leaked foundation language model from Meta, called LLaMA, was leaked in March, and within weeks, people tinkering around on laptops and penny-a-minute servers had added core features like instruction tuning, multiple modalities, and reinforcement learning from human feedback. OpenAI and Google were probably poking around the code, too, but they didn’t — couldn’t — replicate the level of collaboration and experimentation occurring in these smaller communities.

This rapid development of smaller open source projects is challenging the business model of larger companies, like OpenAI and Google. These companies have been pursuing a SaaS model, offering access to high-value software or services through carefully gated APIs.

But customers are starting to question why they need to wade through a large language model like GPT-4, which is the largest and most general-purpose AI model ever created, when all they want to do is match the language of a contract against a couple of hundred others.

The smaller projects are proving that the titanic computation problem that seemed to pose an insurmountable obstacle to challengers is already a relic of a different era of AI development. While bigger isn’t always better, the rapid development of smaller projects shows that smaller can be just as effective, if not more so.

In conclusion, while the dominance of OpenAI and Google is not currently under threat, they will have to keep a close eye on the rapidly growing number of open source projects that are pushing the state of the art and challenging the business models of larger companies. The deep-pocketed companies must adapt and innovate in order to stay ahead of the competition.