Skip to main content

Google is right to fear open source AI

· 4 min read

You wouldn’t expect a behemoth like Google, which has its hands in almost every aspect of technology, to even care about open source and open core technology. But it turns out, it does care. And it’s afraid. Very afraid.

We know about this because an anonymous user on a public Discord server posted what they claimed was an internal Google memo written by Luke Sernau, a senior software engineer at Google. The authenticity of the memo was later verified by consulting firm SemiAnalysis, who obtained permission from the leaker to republish it. You can read the memo at SemiAnalysis’s site.

According to the memo, Google is falling behind the open source community. Despite being engaged in an AI arms race with competitor OpenAI, neither company can keep up with the pace of innovation coming from open source engineers. Furthermore, Google doesn’t have any “secret sauce,” or any special value to give it an edge over open source.

The memo also states that open source models are “faster, more customizable, more private, and pound-for-pound more capable” than Google’s AI offerings. The memo basically sums up the situation as this: Open source AI development is more agile and just plain better than anything the big companies have to offer, despite the volumes of money and resources they’ve thrown into developing their own AI models.

Overall, the memo reveals that Google finally gets what we’ve been trying to say. It realizes that open source AI models have an advantage: an unstoppable wave of innovation driven by rapid iteration and scaling, thanks to input from anyone who wants to have a say.

As the memo puts it, “The barriers to entry for training and experimentation have dropped from a massive research organization to just one person, an evening, and a beefy laptop.” In other words, open source is democratizing AI development, and that’s great news for the final product.

This is exactly how we believe our industry should function. Open source and open core drive innovation. True open source is accessible to everyone, regardless of their background, industry, or location. Developers are free to experiment without the shackles of a company’s agenda, shareholders, or limitations.

When enough people come together to work on technology, the results are bound to push the boundaries of what was previously thought possible. This leads to unexpected innovations that may not have surfaced in a closed environment. The collaborative and distributed nature of open source development allows developers from all corners of the globe to contribute ideas and test code. Open source thrives on rapid iteration, propelling it forward at an astonishing pace.

What’s more, this approach helps combat the biases that can inadvertently seep into technology when only a small, homogenous group is involved. With a larger and more diverse community of developers, the risk of biases creeping into the technology is greatly reduced.

Now, you might assume that Google, with its vast resources, would hold a significant advantage in this space. That’s clearly not the case here. As far as minds go, Google’s payroll is dwarfed by the sheer number of developers contributing to any given open source solution. In fact, the volume of resources may not matter at all when it comes to open source: In the leaked memo, Google freely admits that open source achieves more with fewer resources. According to them, open source does it all—faster, cheaper, and better.

So, what are the broader implications here? Basically, what’s bad for big software giants like Google is ultimately beneficial for developers and end users everywhere. As Google loosens its grip on proprietary technology, opportunities and advantages open up for everyone else, leveling the playing field and fueling innovation.

To sum up:

  1. Google knows it’s behind. Its best hope is to learn from and collaborate with others, rather than trying to operate in a vacuum.
  2. It also realizes that if there are better free, open source options available, people won’t pay for a Google product. So, to monetize or profit from its own AI, it needs to add value to it.
  3. Google says it will redirect its focus on smaller models, because developing giant models is, in fact, slowing it down. It understands that quick iterations will lead to more success.

All of this should ultimately result in better technology at better prices for the end user.

From DeepMake’s perspective, we believe open source and open core are good for technology. We don’t like anything that’s locked down, restricted, or proprietary in any way. We like technology to remain open and available, so everyone benefits.

We’re not crying for Google, and you shouldn’t either. Out here in the wild, no one needs to fear open source — except the Googles of the world.