Poseidon has secured $15 million in seed funding led by a16z Crypto to construct a decentralized knowledge layer designed for synthetic intelligence coaching.
The San Francisco-based full-stack AI knowledge layer stated it goals to deal with the shortage of high-quality, IP-cleared coaching knowledge in AI growth, in response to a Tuesday announcement shared with Cointelegraph.
“LLMs and compute are now not the bottlenecks; it’s high-quality knowledge that’s lacking,” stated Sandeep Chinchali, Poseidon’s chief scientist and in addition chief AI officer at its incubator, Story Protocol.
“Poseidon delivers the IP-cleared, structured real-world knowledge units that AI groups have to construct methods that really carry out in bodily, advanced environments,” he added.
Associated: Decentralized OORT AI knowledge hits high ranks on Google Kaggle
Decentralized pipeline for authorized AI coaching knowledge
Poseidon’s answer depends on decentralized infrastructure to gather and distribute knowledge units legally cleared for business use. The platform integrates Story’s onchain licensing infrastructure to make sure traceability and monetization, permitting knowledge contributors to be paid for his or her work whereas defending builders from IP dangers.
The staff argues that centralized knowledge sourcing fashions can not meet the rising demand for area of interest, high-context knowledge units wanted by next-gen AI fashions, particularly in fields like robotics and spatial computing.
Chris Dixon, founding father of a16z Crypto, described the undertaking as a step towards “a brand new financial basis for the web.” He added that the mannequin rewards creators and suppliers for “offering the various inputs that next-gen clever methods want.”
Poseidon is working with a number of AI labs and plans to make use of the funding to scale its infrastructure. This consists of launching contributor modules, software program growth kits and licensing instruments for builders and knowledge suppliers. Early entry is predicted to start this summer time.
Associated: Daring Applied sciences and My Aion launch $2.5B sensible metropolis AI platform
Poseidon to unravel AI’s knowledge drought
The early wave of AI basis fashions thrived on considerable on-line knowledge, however that period is over, a16z analysts Chris Dixon and Carra Wu stated in a notice shared with Cointegraph.
They famous that simply accessible knowledge units, together with books, web sites and public information, have largely been mined, leaving AI fashions starved for recent, high-quality and legally usable info.
“The problem isn’t simply technical — it’s an issue of coordination. Hundreds of contributors should work collectively in a distributed solution to supply, label and keep the bodily knowledge that next-gen AI wants,” the duo wrote.
They added that no centralized strategy can effectively orchestrate the information creation and curation that’s wanted on the required degree of scale and variety. “A decentralized strategy can remedy this,” they stated.
Journal: AI Eye: Rising numbers of customers are taking LSD with ChatGPT
