Open Source: The Surprising Challenger in the AI Race

The competition among big tech companies in the AI space has been heating up, with some even referring to it as an “AI war.” Microsoft has been leading the charge, thanks to its arsenal of generative AI-powered products, while Google has been making strides, though it has yet to launch a standalone generative AI product or integrate the technology into its flagship Google Search.

However, a senior software engineer at Google believes that the company’s real competition doesn’t come from Microsoft or OpenAI, but from open-source technology.

According to Luke Sernau, founder of Better Engineering and Senior Software Engineer at Google, neither Google nor OpenAI is in a position to win the AI arms race. Instead, he argues that Google’s focus on OpenAI has caused the company to overlook the rapid advancements being made in open-source technology.

Sernau’s remarks were part of a document published in early April on an internal system at Google and have since been shared thousands of times among Google employees. Consulting firm SemiAnalysis later published the document on May 4, and it has been circulating in Silicon Valley ever since.

Open Source: The Underdog That Packs a Punch

When it comes to large language models (LLMs), Meta’s LLaMA is the open-source community’s top choice. Released in February, LLaMA is claimed to outperform GPT-3 in various tasks, including natural language processing and sentiment analysis. It’s also highly customizable, making it more developer-friendly than other LLMs.

However, LLaMA isn’t the only developer-friendly LLM out there, and Sernau is well aware of this. In his words, “While our models still hold a slight edge in terms of quality, the gap is closing astonishingly quickly. Open-source models are faster, more customizable, more private, and pound-for-pound more capable. They are doing things with $100 and 13B params that we struggle with at $10M and 540B. And they are doing so in weeks, not months.”

According to Sernau, Google has no “secret sauce,” and the company’s best hope is to learn from and collaborate with what others are doing outside of Google. He also suggests that clients would be unwilling to pay for restrictive models when high-quality models are available for free.

The Advantages of Open Source

1. Open-source technology is rapidly advancing, and it’s no surprise that it’s becoming a significant challenge in the AI race. Here are some advantages of open-source technology:
2. Faster development: Open-source technology is more developer-friendly, which means that models can be developed and deployed more quickly.
3. Customizability: Open-source models are highly customizable, allowing developers to fine-tune the models to their specific needs.
4. Privacy: Open-source models are often more private than proprietary models since users can review the source code to ensure there are no hidden functionalities.
5. Cost: Open-source models are typically available for free, which makes them an attractive option for those who cannot afford proprietary models.

Conclusion

Open-source technology is proving to be a significant player in the AI race, and it’s clear that companies like Google and Microsoft need to take note. While proprietary models may have a slight edge in terms of quality, open-source models are more developer-friendly, highly customizable, and often more private. As the gap in quality between open-source and proprietary models continues to close, it’s clear that open-source is a force to be reckoned with.