Community matters: Four open source trends in 2024

Open source gained traction with large enterprises over the last two decades, but recently, AI and machine learning shifted even more attention to open source. To build the most technically advanced applications at scale, large companies are increasingly embracing open-source tools, models, and frameworks due to their flexibility and cost advantages.

At Capital One, this is nothing new. We began investing significantly in open source in 2015, embarking on a journey to put open source at the center of our digital transformation. Recently, I was fortunate enough to step into the role of Director of Engineering in Capital One’s Open Source Program Office and I’m inspired every day by how our associates are building the bank of the future using open frameworks. For example, we’ve recently released several ML open source projects including Federated Model Aggregation, Data Profiler, and Synthetic Data.

When we look to the year ahead, the Capital One engineering teams are optimistic about the potential of open source to accelerate our technology transformation as a highly-regulated company. In fact, open-source foundation models from large companies and hundreds of promising community-driven projects are creating an inflection point for the financial services industry. As a result, here are four key open source trends we expect to accelerate in 2024.

As more open source technologies incorporate or embrace ML, we’ll continue to see global movement on legislation that impacts open source driven by rapid change in the AI/ML landscape. These regulations could change how companies use, secure and release any ML-enabled software, and will include clear stipulations around safety, transparency and copyright, to name a few. Here at Capital One, we believe companies in regulated industries can be leaders in finding ways to meet high standards for open source security and reliability, setting benchmarks for compliant and responsible use.

We’ve learned as a community that the open source definition that we use for software isn’t enough to cover the AI space. It’s complicated since there are so many components, from AI architecture and tools, to frameworks, models, and applications. There is a school of thought that supports simplifying the definition of “open source AI” to mean all of the core code must be freely available, but that doesn’t address whether the kernel needs to be transparent and explainable – nor does it take into account training data, weights, and model frameworks. By the end of 2024, we expect to see a standard definition of “open source AI” emerge - and this will be critical for technical adoption and to help form policy. To regulate open source AI and to deploy it widely in regulated industries, users will need to coalesce around a standard definition. The non-profit Open Source Initiative is very active in this area and could lead the charge on creating a standard definition. 

During the uncertain economic times of 2023, some companies cut their investments in open source programs. This year, companies will need to reinvigorate their open source investments. With the aforementioned policy changes, companies will need to hire not just technical staff, but also legal and regulatory experts who can help comply with the upcoming laws. To stay ahead, companies will need to invest time and resources into understanding the new regulations by doing inventories of their open source projects, as both users and contributors. Organizations will expand their in-house open source expertise to ensure compliant use for ML development, as well as invest in building controls and security parameters around their use and consumption of open models.

Open source is first and foremost a community; developers around the world share expertise, contribute code, and learn from one another. To get the most value out of open source, enterprises can’t just “use” it but must also contribute to it. This year, we expect to see companies put a renewed focus on community relations. Internal tech teams will be encouraged to participate in projects and companies will dedicate resources to strengthening their relationships with community members. Enterprises will also build deeper connections to open source foundations to support their AI strategies. 

At Capital One, the Open Source Program Office is keeping all these trends in mind as we build out our 2024 plans. Companies that want to become tech-first leaders will have to embrace the evolving open source AI landscape, as closed-source software is costly and limited. And the best way to benefit from open source is to be in the game—contributing to community-driven projects and helping to build new features. This level of participation can ensure the long-term security and maintainability of enterprise open source projects, making them reliable for use within regulated industries. This year is an exciting time for open source as new regulations and definitions emerge. As users and contributors, we are all part of the community that will put open source at the center of many large enterprises.