Features Hub Opinion

How open source is accelerating innovation in AI

Wed 11 Sep 2019 | Carmine Rimi

By eradicating barriers like high licensing fees and talent scarcity, open source is accelerating the pace of AI innovation, writes Carmine Rimi

No other technology has captured the world’s imagination quite like AI, and there is perhaps no other that has been so disruptive. AI has already transformed the lives of people and businesses and will continue to do so in endless ways as more startups uncover its potential. According to a recent study, venture capital funding for AI startups in the UK increased by more than 200 percent last year, while a Stanford University study observed a 14-times increase in the number of AI startups worldwide in the last two years.

As AI innovations and applications continue to advance, a growing number of technology companies including Microsoft, Salesforce and Uber have started to open-source their research and projects. These companies, which have substantial Research & Development (R&D) capabilities, are investing in or “giving back” to the open source community to reinforce the creation and improvement of AI & Machine Learning (ML) algorithms.

Consequently, open source software has become crucial to driving fast, reliable and secure AI development. But why did the tech industry turn to open source to power AI, and how will it influence its development moving forward?

Open source goes mainstream

The first open source program, Netscape Navigator, was launched in 1998. Open source software has played a significant role in technological progress ever since. The “strategy of openness” Netscape chose was to emphasise the business potential of sharing the software’s source code and one that drew inspiration from academia — if researchers kept research methods hidden, innovation would develop at a snail’s pace, limiting scientific progress.

As more people woke up to the wide applicability of these cooperative principles, they soon left the academic petri dish and spurred a gear-change in IT innovation, one with a holistic vision that benefited more stakeholders. With developers on a mission to embed AI into IT operations across industries worldwide, secure and scalable software frameworks are vital.

The enormous computing power required for the development of AI and ML models is usually the reason behind high costs of R&D and the amount of data required to build and train advanced models is usually a challenge for developers.

Furthermore, a growing skills gap among software engineers remains a big challenge for the industry. According to a recent study by Tencent, despite the millions of AI jobs advertised globally, there are only 300,000 professionals able to fill them.

“Knowledge sharing between IT teams and companies across industries allows developers to have access to trusted, secure and easy-to-deploy solutions”

Compared to the significant fees that come with licensed software, open source software enables developers to use publicly available frameworks, data sets, workflows, and software models, thereby reduces training costs. Moreover, the open source community serves as an extra layer of security and is always monitoring projects for vulnerabilities.

We need to look no further than Kubernetes, the open source platform for automated deployment and management of containerised applications, including AI and ML workloads. Any analysis of the relatively smooth and meteoric rise of the container orchestrator is incomplete without mention of the strong and ongoing community effort that sustains it.

Key ingredients for AI: Openness and trust

So, why does open source play such a disruptive role in the development of new technologies?

Open source fully democratises the R&D process, thereby allowing developers to build cheaper, faster, more flexible and secure deployments and with the help and support from a large number of contributors, open development also helps to accelerate the adoption of many key frameworks and software solutions.

Tech companies are demonstrating a commitment to contributing to and supporting the AI open source community by making AI and ML frameworks accessible to everyone.

The latest big contribution to the AI and ML open source resource library is from Facebook, with its Deep Focus (the AI rendering systems data and code) becoming publicly available at the end of last year. Google is another major contributor. Starting with Tensorflow — its popular machine learning framework used by companies including Airbnb, Uber, SAP, eBay — the company has been opening up its research to the public.

Amazon has also recently opened up some of its internal ML courses to developers. The company sees this as an opportunity to not only hire more engineers but also accelerate machine learning growth. It’s clear that with such a heavy investment in “openness of AI” from established tech companies, AI development will continue to accelerate over the next few years.

By eradicating barriers such as high licensing fees and talent scarcity, open source AI and ML can accomplish its goals more easily and more quickly. In particular, knowledge sharing between IT teams and companies across industries allows developers to have access to trusted, secure and easy-to-deploy solutions, moving entire industries forward by making AI more accessible.