Why the Anti-Open Sourcers Have AI All Wrong

Critics of open-source AI raise concerns about the safety and responsibility of releasing powerful AI tools so broadly. The argument relies on the perceived dangers of putting advanced AI technologies into potentially unqualified or malicious hands. Skeptics worry that without proper safeguards, open AI can lead to unintended consequences, ranging from the spread of misinformation to the creation of more sophisticated cyber threats.

But they’ve got it all wrong. In this article, I will demonstrate why their concerns are unfounded and how open source principles, with their focus on transparency and collaboration, consistently win over the restricted nature of walled gardens and proprietary secrets.

To understand the future, look to the past

The history of tech’s evolution shows time and time again how open source triumphs over all the rest. Apache and Nginx outlasted and won over IIS, due to their adaptability and community-driven enhancements. In web content, proprietary tech such as Flash and Silverlight were dropped in favor of HTML5 and JavaScript—open standards that now define our web experiences.

At the age of 10 I started my fascination with Linux (specifically Red Hat Linux 5): The concept of having access to the source code of the entire operating system I was using, having the ability to fully understand and customize every aspect of it felt very empowering. Not long after I installed my first Linux distribution, did I also understand the much larger world of GNU and other open source and free software tools. These experiences essentially made my core tech philosophy: Open source is the foundation for sustainable innovation.

Nowadays even large corporations understand that Open Source is the only way to achieve mass adoption. The most commonly used IDE (code editor), VSCode, is fully open sourced and is sponsored and developed by Microsoft. Tell that to someone in 1999 and they would’ve thought you were pulling an elaborate prank. Back then, the concept of a tech giant embracing and actively fostering an open source environment was almost unthinkable.

This shift underscores a major change in the industry’s approach towards innovation—a realization that opening the gates to community contribution can push technology forward more effectively than the restrictive control of proprietary development ever could.

In the DevOps arena, open source tools like Docker, Kubernetes, and Terraform are critical to the way we build and deploy applications. Docker packages software into standardized units for deployment, Kubernetes orchestrates and scales these containers. Terraform allows engineers to use code to set up and maintain the infrastructure. These tools have become critical for efficient development processes, illustrating the innovative and community-driven spirit of open source.

Why Open Source for AI?

With products like ChatGPT, Midjourney, and Notion AI, we’re experiencing increased efficiency only comparable to the internet revolution. ChatGPT is redefining user interaction by providing conversational AI that can answer queries, write code, or even write simple unit tests, while Midjourney is enabling users to bring their visual ideas to life without any PhotoShop skills or a background in design. Notion AI complements by streamlining organizational tasks, from automating to-do lists to synthesizing complex documents, effectively elevating productivity across the board.

However, these products, like the majority of AI products currently in the market, are fully centralized. This means that if one of these companies shuts down, all your personal training data and access will shut down with it. As we increase our dependency on these services, we need to be wary of the fact that we’re potentially putting all our eggs in one basket.

Picture the repercussions of AWS going offline—now imagine what would happen if ChatGPT drastically changes or pivots in a few years' time. Only a self-hosted AI revolution could mitigate that level of disruption.

This is precisely the reason why we should support and foster a more federated, decentralized, open source approach to AI products: To make sure we have ever-green solutions that outlast individual products and companies. This would ensure that even if the product itself is no longer available, your training data and ability to run that same product locally or on your own cloud remains unchanged.

There are four main reasons why open source AI benefits everyone, not only the end users:

• Interoperability: open source software often supports cross-platform compatibility, a critical foundation for diverse tech ecosystems.
• Community Support: Global contributions mean rapid development, diverse perspectives, and fast troubleshooting.
• Vendor Neutrality: With open source, there’s no lock-in effect. Users maintain control, choosing how and where to deploy.
• High Extensibility: Imagine Firefox or WordPress without their extensions. Open source allows for massive modularity, and gives spaces for companies to continue on their core business, like WordPress’ focus on blogging, while the community develops secondary businesses, like WooCommerce’s ecommerce platform.

Open Source always wins, eventually

As we witness the true beginning of the AI revolution, the time is right to ensure this is a long-lasting truly innovative technological evolution by building the right foundation based on open source technologies. This is the only path forward to avoid the pitfalls of historically proprietary technology, such as Flash or Silverlight, and choose to build the next HTML or Linux.

The AI revolution, propelled by open source, promises more than just technological advancements. It offers a future where self-hosted AI solutions are as common and essential as the web servers and development tools we rely on today. My 10-year-old self was onto something: Open Source isn't an option, it's the foundation for an innovative AI future.

Asaf Zamir is a CTO and Cloud Consultant on the A.Team builder network.