There is a new crypto project that aims to build a decentralized network for confidential AI output verification. It is called Mira, it was launched last year and it is still in the test network phase.
The Crypto network was aimed at Mira AI
The project was initiated by three AI experts from Aroha Labs, Ninad Naik, Sidhartha Doddipalli and Karan Sirdesai.
In particular, Ninad Naik has previously been the AI leader with Uber and Amazon, and on Mira He holds the position of Chief Product Officer. He has extensive experience in making AI platforms.
At Uber he led the development of the most important market product for the worldwide delivery activities of the company, while he at Mira He leads product development and research to enable developers and companies to use artificial intelligence in new and impactful ways.
Mira’s mission is to build an open infrastructure for the Supply Chain Ai.
The aim is to determine new standards for the way AI is being developed, used and maintained, with a certain focus on minimizing complexities for web2 and web3 developers.
The project is still in the development phase and is open to developers, makers and researchers who want to investigate the potential of the infrastructure and participate in the ecosystem.
How does the Crypto network work for AI from Mira?
The starting point is it Plauselele AI output.
In fact, artificial intelligence often produces output that contain incorrect information due to the probabilistic nature of technologies based on neural networks.
The MIRA network wants to be a platform where the outputs generated by AI can be verified with decentralized consensus, that is, confidentless.
In other words, rather than having to rely on verifies who have to place their trust, the verifiers will be reliable and the verification can be done by someone without having to trust someone, except the software of their platform.
And so the outputs of AI are converted into statements that can be independently verified by the users themselves, so that different AI models can also jointly determine the validity of each statement.
The node operators of the network that perform this inference-based verifications are economically stimulated by a hybrid proof-of-work/proof-of-stake mechanism, to make them carry out a fair verification.
“` HTML
The problem and the solution proposed by Mira
“ `
Today’s AI systems excel in generating creative and plausible outputs, but they are unable to offer flawless outputs in a verifiable and reliable way.
That is why verification currently requires human supervision, while the decentralized verification network
Mira wants to achieve the same result, if not better, with a new combination of content transformation and distributed consensus made possible by crypto-economic stimuli. This should make manipulation both technically and economically impractical.
In addition to verification, the vision of the MIRA project is to achieve a synthetic foundation model that integrates verification directly into the generation process.
Such a simplified approach would eliminate the distinction between generation and verification, so that output already produces errors.
In addition, the verification over a decentralized network of stimulated operators creates an infrastructure that is inherently resistant to centralized control.
In the white paper they write:
“This represents fundamental progress: by letting artificial intelligence systems work without human supervision, we lay the foundation for a real artificial intelligence, a crucial step in the direction of unlocking the transformative potential of artificial intelligence in society”.
“` HTML
The sign of the MIRA project
“ `
The MIRA network is currently still in TestNet. Because the Maintnet does not yet exist, the cryptocurrency does not yet exist.
Nevertheless, there are already several tokens with this name or with this ticker on the crypto markets, but they have nothing to do with the Mira Network project of Sirdesai, Doddipalli and Naik.
At the moment there is not even a public route map to know when it will be launched, although the first phase of fundraising ended in July last year with a collection of 9 million dollars.
The most important investors are Bitkraft Ventures and Framework VenturesBut Accel, Mechanism Capital, Folius Ventures and AJ Scaramucci’s Salt Fund Also participated in fundraising.