There are several
characteristics of blockchain (or more generally, distributed ledger)
technology that are of potential interest for artificial intelligence
applications. The decentralized nature of blockchains allows for convenient
sharing of AI training data, training processes and of pretrained AI models.
While in the past, traditional databases and repositories have been used for
sharing data, the trustless nature of blockchains, along with the transparency,
immutability and auditability provided via cryptographic provenance allow for
secure sharing, with improved trust on the AI models and the data they work
with. Blockchains can also serve as a global registry for AI data and models,
with appropriate permissions and access controls enforced.
This paper provides a brief background on blockchains and AI applications, how they can benefit from each other and also debunks some of the myths on the synergy between the two.
Read more here: