Decentralized AI on Verge of Historic Inflection Point

The development of AI technology is following a similar trajectory to that of Bitcoin in its early days. Both have reached important milestones, and their growth has been exponential.

Bitcoin hit an important milestone when it mined 100,000 blocks, with the price also near $1 per coin at the time. The network was maturing from proof of concept to something legitimate, but there were still hurdles to overcome, such as user-friendly exchanges and conversion options.

The number of miners active on the Bitcoin network was starting to grow, which is essential for maintaining the security of the network. They validate transactions, help distribute new coins into circulation, and ensure transactions remain permanent and tamper-proof.

The growth of AI technology is also experiencing an inflection point, similar to that of Bitcoin. Large Language Models (LLMs) took off with OpenAI's first model, GPT-1, released in 2018. The parameters used for each model exploded in the years that followed, with GPT-3 being a major turning point.

Decentralized AI is also reaching an inflection point, where many computers across the globe train a model. This technology must coordinate separate tasks and blend them together into one cohesive result, which is a complex challenge. Researchers from Macrocosmos are pre-training a 16-billion-parameter model using a mix of graphics cards, similar to what Pluralis Research did last year with their Node-0 project.

The growth in decentralized AI infrastructure has the potential to create new verticals within the AI space for everyday individuals. This includes renting the network for pretraining runs and creating more niche and specialized models.