Open-source AI promises to address a lot of the ethical concerns around AI as well as helping to drive much greater levels of innovation than closed-source models from big tech companies. Here’s how open-source could eventually outpace big tech, ushering in a new era of decentralised artificial intelligence. Chris Price reports.

It’s clear that the divide between open-source and closed-source AI looks likely to shape tech industry dynamics over the coming months and years. Driven in part by Chinese company Deepseek’s much-publicised release recently, open-source AI is starting to attract attention, particularly from enterprises attracted by its lower costs and greater flexibility. However, closed-source models still remain by far the most popular AI solutions available to businesses right now.

But what exactly are the differences and why should developers and enterprises consider going open, rather than closed-source, in the long term? Firstly, closed-source AI companies such as Open AI – producer of industry leader ChatGPT4 – operate largely behind closed doors. Their algorithms, training data and model parameters are kept secret, accessible only to the company that developed them. And while this approach has clearly proved extremely lucrative so far (Open AI has as estimated valuation of $157 billion according to research company CB Insights), it has also attracted much criticism within the tech community.

For example, in a closed-model, power and control over AI systems is concentrated largely in the hands of a few large corporations, such as OpenAI, Anthropic and Google, potentially stifling competition from smaller rivals. Furthermore, access to these models is often restricted and expensive, which can prove a barrier for smaller organisations. Another problem is around their lack of transparency. The opaque nature of closed-source AI models can make it difficult to understand how they arrived at their decision, which in turn raises suspicions and concerns about their bias and accountability.

Embracing collaboration and innovation

Conversely, open-source AI embraces transparency and collaboration. The code, data and models are freely available for anyone to access, modify and distribute, thereby fostering a much more community-driven approach to development. This can help to create more of a level playing field, allowing smaller – often more innovative companies to compete in the AI revolution which they previously wouldn’t have been able to do.

Indeed, it’s an approach that has so far been favoured by Meta for its Llama LLMs (Large Language Models). As Meta CEO Mark Zuckerberg stated in a blog post to launch its Llama 3.1 LLM last year: “To ensure that we have access to the best technology and aren’t locked into a closed ecosystem over the long term, Llama needs to develop into a full ecosystem of tools, efficiency improvements, silicon optimisations, and other integrations,” he wrote. “If we were the only company using Llama, this ecosystem wouldn’t develop and we’d fare no better than the closed variants of Unix.”

Stable Diffusion is another example of a community-driven, open-source AI platform. A powerful text-to-image AI tool, its core code was made publicly available to developers and enterprises, allowing anyone to download, use, and modify it. This immediately lowered the barrier to entry for individuals and smaller teams who previously lacked the resources to develop such sophisticated AI. It also helped to spark a massive wave of community-driven development including integration with image editing programs and game development tools.

Helping to build trust

Nor is innovation the only advantage of an open-source AI model. Decentralisation is another key benefit. By distributing access to AI technology, it can help to prevent the concentration of power in the hands of a few corporations. For futurist Eric Bravick, CEO and Founder of The Lifted Initiative, decentralised open-sourced models are vital to establish transparency and build trust. “In a decentralised system, the origin of data is as critical as the data itself—establishing ownership that consumers can trust and control,” he says.

Furthermore, decentralised AI is essential to help prevent misuse of technology. “History teaches us that the concentration of power is antithetical to human liberty,” he adds. “I believe decentralised AI offers the best defence against the potential abuses of this technology.”

Founded by Web2 and Web3 experts with experience at Google, Apple, Microsoft, Facebook and many others, The Lifted Initiative is building a decentralised physical infrastructure for its customers through its Manifest Network. This comprises a suite of tools for enterprises including a distributed network of GPUs (Graphical Processing Units) for AI processing as well as secure decentralised storage and scaleable, on-demand compute power.

Look to the future

Although closed-sourced AI currently dominates the market, the long-term potential for open-source AI looks extremely encouraging. The collaborative nature of open-source development allows for faster iteration and innovation while the decentralised nature of the technology helps to foster greater transparency.

Rather than hiding behind closed doors, open-source platforms are open to scrutiny, enabling researchers and developers to examine the code and data used to train models and identify any potential biases. Furthermore, open-source AI encourages the democratisation of technology, preventing control from being held in the hands of a few large, increasingly powerful big tech companies.

Of course, challenges still remain for open-source platforms, not least around funding as well as maintaining both quality control and high levels of security. However, as open-source continues to build momentum it may well redefine how AI innovation happens – setting a new pace that big tech’s slower, proprietary models will struggle to match.

The post We’re in a Decentralised AI Revolution appeared first on Compare the Cloud.