Early ICP Builder Arjaan Buijk Recognised for Advancing ‘Proof of AI Work’

A post circulating within the Internet Computer community has drawn attention to Arjaan Buijk, described as one of the ecosystem’s earliest and most dedicated builders.

Buijk discovered the Internet Computer Protocol roughly 1.6 months after its genesis and quickly became deeply involved. Among his early technical achievements was getting the first large language model running inside a canister, a milestone for on-chain AI experimentation within the network.

He later co-founded ONICAI alongside Patrick Friedrich, known online as @thepatnorris. Together, they formalised and popularised the idea of “Proof of AI Work” for decentralised AI networks.

The concept aims to address a growing concern in blockchain and AI circles: how to verify that artificial intelligence systems are genuinely performing useful computation. Rather than relying on mechanisms such as Proof of Stake or Proof of Work, Proof of AI Work proposes that AI models should earn value only when they can demonstrate that real inference or training has taken place.

At its core, the idea combines cryptographic verification with economic incentives. It seeks to ensure that an AI model actually ran inference or training, that its output is reproducible or verifiable, that the computation involved measurable cost or effort, and that the result can be checked without rerunning the entire task.

Supporters argue this framework could help prevent fake AI outputs, deter sybil models posing as independent actors, and reduce the risk of centralised APIs claiming computational effort without evidence.

The proposal has gained attention as decentralised AI projects look for ways to build trust into their systems at a protocol level. While still an emerging concept, Proof of AI Work reflects a broader push within the ICP ecosystem to align technical innovation with verifiable accountability.