Paris, FRANCE – With over double the number of attendees compared to the previous year’s EthCC, and with over 250 side events, almost every event was oversubscribed. The mood at the Ethereum focused conference last week was elated.

The blue skies and breezy weather of Paris, as well as Ripple’s partial victory in its case with the U.S. Securities and Exchange Commission, didn’t hurt either. In the past few weeks, some of the world’s biggest asset managers have announced applications for exchange-traded bitcoin (BTC). This is not directly related to the Ethereum Blockchain per se but a positive development nonetheless.

The rise of artificial Intelligence or AI was another hot topic, with many talks and side events being organized.

The majority of investors and entrepreneurs I spoke with were optimistic about the intersection between the two fields, even though they differed on how it should be done.

AI in Web 3 Finance

The best use of AI in crypto markets and communities is seen by one camp as immediate, practical. Ken Timsit, from Cronos labs (the Layer 1 blockchain for exchange, sees AI, as an instrument to increase productivity in the crypto financial markets and increase the dollar value managed by crypto firms.

It might seem like a simple matter of putting chatbots on exchanges, but bringing analytical abilities to crypto traders and institutional investors could be a powerful tool. AI has been used by high-frequency traders and market-makers for some time, but they have an advantage in their own technology.

Upshot has been working on this for three years. According to them, NFTs can bring trillions of dollars worth of assets into Web 3 such as luxury goods, art, insurance, and real estate. These non-fungible assets require a lot more human power to continue rolling. AI may be the key to understanding pricing mechanisms and clearing markets.

Better AI with Cryptography

Zero-knowledge (ZK) proofs have become a popular term in cryptography over the last year. They are a type of process that allows you to verify a statement’s validity without divulging any additional information.

ZK can be used for machine learning in order to verify which model was used to produce a certain result in a ZKML mashup, or what datasets have been used in training data. Blockchain could create a trust system decentralized for AI.

According to Eli Ben-Sasson, Co-Founder and president of StarkWare and a key figure in the development ZKs, Blockchains are particularly good at social co-ordination, while ZKs provide privacy, which machine learning does not need.

A venture capitalist stated that keeping track of data used to train AI systems is the best application for blockchain. This is a computationally demanding task. Another crucial task is to keep track of the algorithms. This may be even more critical.

Computing power, performance and usability are still a bottleneck for ZK proofs. These problems are being addressed by companies like Manta Network, Fabric Cryptography, and others.

Blockchains are also being used by other projects to coordinate and pool computational power, which can then be used for training and running machine-learning models.

You can have a great discussion in Paris with croissants and Champagne on rooftops that overlook the Eiffel Tower.

It remains to be seen what exactly the use cases are.

Bradley Keoun, Parikshit Mihra and Bradley Keoun edited the book.