A Singapore-based organization aims to launch a non-profit community that is filled with people who are token-incentivized to validate AI-generated metadata.
Rex Wong said that iVideoSmart which provides white-label video and ad tech platforms has an addressable user base of 500 million globally. As of today 300 of these customers are validating the metadata and further textual descriptions that are generated by AI for translating, transcribing and ensuring the brand safety of videos.
He further continued that the current human workers are being paid for their work, however they are not able to scale for large applications like video search engine which would be required to generate accurate metadata for millions of videos. That is why to achieve this task Avion has announced that it would be creating a blockchain based ecosystem with tokens which can be used to incentivize users.
The tokens generated through Ethereum blockchain protocol will be available in October. These tokens can be redeemed on open token exchanges or can be used to buy content or other products offered in iVideoSmart platform or by Avion community members. These tokens can also be earned and used by content providers, AI providers, advertisers and other parties involved in the ecosystem.
Governed by a non-profit
Once the tokens are added, a video search engine which is powered by the quality of the videos’s metadata will be launched. AI will make the first attempt at textually describing the videos after this the humans will validate or modify AI’s work.
Wong said that this Open Community Video Search Engine will be governed by a non-profit foundation which is controlled by voting members of the Aivon community. Although the advertising applications created using this protocol will be for profit.
The open-source protocol and the community of incentivized human validators can be used by other organizations and companies to create their own search engines or other decentralized apps like content rights management.
Wong acknowledged that AI is rapidly evolving and in the near future it may get above 90 percent mark for meta-tagging accuracy. And when this happens humans wont be required for correcting or adding a text; however they would play their part in confirming the AI’s accuracy.
Please refer to the full article for more information.