Anyone interested in tech and the future of the web has most likely already done the fun mental exercise of taking the models we know and trying to imagine them from the perspective of web3. What would be the consequences of their decentralization? How much sense would a token make? Fungible or non-fungible? What would be the business model? etc. With this in mind, it’s thrilling to re-imagine Stripe, Airbnb, Uber, WhatsApp, Facebook, Google ...

I’ve done this little exercise on a model I know all too well, the image bank. In 2004, I created Fotolia, an image bank with user-generated content. While traditional image banks were only open to professional photographers, Fotolia allowed any amateur photographer to put up their photos for sale. It’s fascinating to try and imagine what Fotolia would look like from a 3.0 perspective. Image banks will become decentralized marketplaces in which each image will correspond to an NFT available in one or more copies. When a customer purchases an image, they will therefore purchase an associated NFT contract (smart contract) which will define the conditions of use and the photographer’s payment.

We can go one step further in this exercice to try and imagine the limits of a model that would be fully decentralized.

At Fotolia, what was the nature of our role and why did it justify a fee? I’ll sum it up in three points: to select the best images, guarantee their legality and make them findable. A huge portion of our costs and energy were spent performing these three functions. How can this be addressed in the era of decentralization?

Two solutions exist, collective intelligence using DAOs under the condition of finding a good governance model, or artificial intelligence. In the case of Fotolia, the recent deep learning technologies may now produce decent results for selecting the best images, identifying those that may violate the law, and indexing them (find related keywords).

The decentralization of services will be likely accompanied by the development of artificial intelligence models. Doing so without a trusted third party certainly means replacing some humans with machines for specific functions like enrichment, arbitration, curation or moderation.

Web3 will either come about through artificial intelligence, or not.