A Technical Product Manager and ICT expert, Paul Joe, has revealed that the global future of artificial intelligence is taking shape along centralised corporate control and decentralised open collaboration.

In his paper, The Future of AI: Centralised vs. Decentralised Approaches, Joe examined how these competing models could define the next era of AI development, with implications for access, trust, resilience, and innovation.

According to him, centralised AI, dominated by major firms like OpenAI, Google DeepMind, and Anthropic, has driven rapid advancements through massive computing power and tightly controlled systems.

Joe argued that even though these companies develop and maintain powerful models such as GPT-4 and Claude behind closed doors, offering limited access through paid APIs, he warned that such control concentrates power and limits transparency.

The ICT expert emphasised that decentralised AI is gaining momentum through open-source models and community-led networks, while projects like Meta’s LLaMA and Stability AI’s Stable Diffusion have released models publicly, enabling global collaboration. Joe added that blockchain-powered initiatives such as Bittensor and Gensyn go further, allowing anyone to contribute computing resources and help train models on distributed networks.

The paper also highlighted key comparisons between centralised AI, which often limits access to large organisations with funding, while decentralised models lower barriers for startups, researchers, and public-sector players.

Joe also pointed out that black-box models run by a few firms can fuel public suspicion, especially around bias and data use. In contrast, open and decentralised systems enable scrutiny and community oversight.

He noted that public trust grows when systems are visible and collectively governed, citing polls that show strong support for more transparent AI development.

He added that resilience is another advantage of decentralisation, while centralised services face single points of failure, making them vulnerable to outages or censorship.

Joe said distributed systems, however, are more robust and able to function even if parts of the network go offline, which makes decentralised AI critical for infrastructure and public service applications.

He noted that innovation thrives in open environments. “While large companies invest billions in AI, community-led ecosystems foster competition and experimentation. A recent survey found that 75% of Americans believe decentralised AI will drive greater progress than models controlled by a few corporations.”

Joe envisioned a hybrid AI landscape over the next three to five years, pointing out that centralised providers are likely to maintain their lead in large-scale models but may be pressured to adopt more transparency. Meanwhile, decentralised platforms will mature into real-world tools for governments, educators, and enterprises seeking flexibility and control.

He predicted a balanced AI ecosystem similar to the internet, where powerful hubs coexist with a wide web of smaller and collaborative nodes.

“If developed responsibly, AI can become more accessible, trustworthy, and innovative — not just for the privileged few, but for society as a whole.”