Francis Viviers, Senior Solutions Architect at Red Hat.

Advances in artificial intelligence (AI) and machine learning (ML) are nothing short of a technological revolution in the making. There are a tremendous number of applications for AI and ML (many of which we are only still only discovering) that are helping governments and businesses automate repetitive tasks and solve complex problems across almost every industry. And as the amount of data we generate every day continues to increase exponentially, AI is helping us make sense of it in ways we could never do manually.

At the same time, a structural shift is happening in the world of AI. In the past, the leading AI platforms were dominated by a small handful of tech companies that held all the keys to innovation.

More recently, however, companies such as Google, Microsoft, Facebook, IBM, and Amazon have begun democratising access to their advanced AI technologies by building open source AI platforms or making existing ones available publicly. With so many well-established companies investing in the ‘openness of AI’, this has set a clear trajectory for the development of the future of AI.

The openness of AI comes as a significant benefit for businesses of all sizes looking to realise its value. Organisations can freely reuse leading AI technologies under open source licences while leveraging the support of a diverse community of developers. And because open source uses a global network for feedback, collaboration is also accelerating the speed at which AI algorithms are being created and improved. More businesses are drawing value from AI and ML the open source way; here are some of the reasons why.

Democratising AI

The open source AI community encourages a culture of sharing ideas and building on existing technologies. This gives companies much greater freedom to innovate with AI – with none of the adoption barriers such as high licensing fees and limited in-house AI expertise. A proprietary approach to AI, on the other hand, doesn’t always create feedback loops that improve development. A closed approach often stifles development and leaves developers with skill sets that are not portable across employers, projects, or even cloud environments.

Of course, proprietary AI solutions still have their place, such as delivering vertically integrated solutions for very specific use cases. But open source AI is levelling the playing field and enabling every organisation to tap into free, readily available, and industry-leading AI solutions in a way that is easy to adopt at scale, more integrated across different environments, and does not result in vendor lock-in. Some tech giants have also realised that the industry shift towards open source AI benefits them too. They are opening up their AI code bases to ensure that competitors with proprietary solutions will struggle to keep up with the pace of open innovation.

Data-driven business and AI

In the era of big data, organisations experience significant challenges when it comes to understanding and using data to their advantage. Luckily, that is where AI excels. But AI models can be expensive to develop and require massive amounts of data to train. With an open source approach, organisations do not have to spend their budgets on AI software they haven’t built and cannot customize, nor do they have to build everything from scratch.

To take advantage of AI the open source way, organisations need only start by defining what they wish to achieve with AI and understand what kinds of data they have at their disposal. They can then choose between a multitude of best-practise open source code bases, and tailor an AI or ML solution to fit their specific use case. Businesses that need to build applications with highly customised data cannot always simply buy a product off the shelf. Open source allows them to become co-inventors of software – not just buyers of it – and everyone benefits from the success of others.

Breaking down the silos

With the growing variety and complexity of today’s IT environments, many organisations face the technical challenge of being able to access and enrich their data when it exists in different proprietary formats. Companies need consistent and interoperable data platforms to unlock the full potential of AI and ML, or they risk suffering from suboptimal application performance, poor data exchange, or halted integration and development due to cross-compatibility issues.

Organisations can benefit from a data strategy built on open standards as it protects them from vendor lock-in, allows for seamless integration between different types of data, and gives them the flexibility to support current and future data formats. This enables them to improve their technological capabilities without proprietary constraints, while staying on the cutting edge of rapidly evolving AI and ML technologies.

With vast amounts of data traffic moving to the cloud, a more interoperable open source approach will become increasingly important. In 2017, 90% of the global public cloud workload was run on Linux. Furthermore, much of the existing AI ecosystem is already built on open source. Python, for instance, is the open source programming language behind most of today’s AI applications, and the same is true for leading open source AI and ML platforms such as Apache Spark, TensorFlow, PyTorch and IBM Watson and many more.

An open future for AI

The worlds of AI and open source software are steadily merging. The ever-growing community of AI developers is working to make algorithms smarter, product cycles shorter, and AI software more reliable, accessible, and integrated. There’s no doubt: the future of AI is not only exciting, but also more open. As more businesses – big or small – begin to realise the untapped potential of AI and its growing variety of applications, more of them will begin to embrace the benefits of taking an open source approach.