Open-source AI is a system that is free to use, study, change, and share. It is developed in a collaborative way, with the source code, datasets, and model parameters available to the public
Prime Minister Narendra Modi while speaking at the Artificial Intelligence (AI) Action Summit in Paris recently, emphasised the need for AI models to be built on an open-source framework. According to him, this initiative will enhance trust and transparency in the model. It will also democratise technology and ensure the development of people-centric applications.
PM Modi’s urge to adopt open source AI comes at a time when China’s open source model DeepSeek has shaken up the AI industry. DeepSeek-V3 is one of the best performing open source AI models in the world. The Chinese start-up’s low-cost models are at par with its Western counterparts like OpenAI and Google’s closed source models.
Let's suppose there is a fully functional AI model called ‘ABC’. It is developed using a dataset, on a fixed code and trained on some parameters. For ABC to be considered as open source, it should be made available (under terms) in a way that grants anyone the freedom to use, study, modify and share the model as a whole and its structural elements for any purpose and without having to ask for permission.
Open-source AI is a system that is free to use, study, change, and share. It is developed in a collaborative way, with the source code, datasets, and model parameters available to the public. The purpose of an open source AI is to promote and cultivate transparency, innovation and community support.
Apurv Agrawal, CEO & Co-Founder, Squadstack said, “The real power of open-source AI isn't just in cost savings or accessibility - it's in how it's democratizing innovation at the architecture level. While proprietary models excel at general intelligence, open-source creates opportunities for specialized intelligence. Think of it like the early days of Linux - what made it revolutionary wasn't the operating system itself, but how it enabled developers to build specialized solutions for unique contexts.”
Being open source seems like an act of charity, building a system and providing it to the world for free. However, it does have some hidden benefits for the start-up as well.
If you have a model that is strong enough to be used for multiple use cases, making it open source can market it for free. Start-ups will build products on top of your model and promote it, thereby promoting your model as well.
"Start-ups today embrace open source despite its challenges because it fuels innovation like never before. And in many cases, builds trust and drives adoption. The risk of competitors copying their code definitely exists, but the overall benefits outweigh it. Start-ups that use open-sourced inspiration tap into a global developer community, accelerating product improvements at a fraction of traditional R&D costs," said Abhilash Madabhushi, Founder, Consuma.
Also, when you open source a model, several players jump into the segment to develop similar technology. These players might look like competition but they will research and invest in the segment to replicate your model which will expand the market for it, thereby increasing your pie.
Talking about start-ups using open source models to enhance their products and services, Krishna Vij, VP, TeamLease Digital said, “Open-source AI is a game-changer, especially for startups. It reduces the cost of innovation since companies don’t have to build models from scratch. They can access pre-trained models, fine-tune them, and focus on solving real-world problems rather than reinventing the wheel. This speeds up time to market and encourages collaboration.”
If the developer or owner of the supposed fully functional AI model ‘ABC’, prohibits the public from accessing its source code, training data, and model architecture, it will be called a closed source AI.
Closed source AI is AI software that keeps its models, training data, underlying code and model architecture confidential. This gives the company that owns the software more control over the system. It protects the research and development of AI technologies and can be better at specific tasks than open source counterparts but loses out on the benefits of public oversight.
AI startups often choose to be closed source to protect their intellectual property, maintain control over their technology, ensure data privacy, and potentially generate more revenue by offering exclusive access to their advanced AI models, especially when working in highly competitive sectors. This allows them to prioritise dedicated support and fine-tune their algorithms without exposing the underlying code.
Contrary to its name, OpenAI is closed source. It keeps the information about its advanced models confidential and provides access to these models via API (Application Programming Interface).
Several users have wondered why ChatGPT isn’t open-source. Based on available information in OpenAI’s developer community, OpenAI appears to be concerned about the potential risks associated with making ChatGPT fully open-source. By keeping the code private, they aim to prevent malicious usage and other unintended consequences that could arise if it were to fall into the wrong hands.
"India has an excellent opportunity to create its own language models, but the real advantage is that applications can be built to tackle specific challenges in different industries. By enhancing existing models with localized intelligence, startups can accelerate innovation and build solutions tailored specifically to India’s diverse workforce," said Surinder Bhagat, Founder and CEO, Gigin AI.