<aside> 🏗️ Algovera will be funding squads to build their own projects. We are building tools to make the process of creating AI projects easier.
</aside>
Hi, we’re Algovera. We envision a world where independent AI developers have the freedom to work on projects that they’re passionate about. These AI developers would keep ownership of their ideas and inventions, receive rewards based on their contributions and acquire passive income based on the value generated by their creations. With this in mind, we are launching the Algovera grants program for projects that combine AI and Web3. For the first round, $5000 worth of grants are on offer. Initially, the maximum grant for each individual project is $1000 (e.g. to fund one month of part-time work).
On our journey to date, we’ve realised that achieving our vision requires adopting several new technologies. In general, it needs to be easier for independent developers to launch useful AI apps and we’ve identified a few things that we think might help. For example, datasets need to be more accessible and permit use for commercial purposes. The privacy of the data owners should be better preserved using maturing private AI technologies. It should also be easier to coordinate and collaborate among distributed teams using tools such as shared treasuries, governance mechanisms and project tracking. Marketplace infrastructure (such as Ocean Protocol) should be available for quickly advertising and monetising processed data, algorithms and apps. And finally, cheap storage and compute resources should be more widely available for training models and storing data and weights.
Practices that include these new tools result in a workflow that is quite different to the ones that AI teams are used to. To better map this out, we organised a series of AI x Web3 hacking sessions, where we went through the whole process of developing a toy algorithm using some of these new technologies in a shared GitHub repo. For the reasons outlined in this blog post, we chose a generative algorithm for creating art NFTs. We have since launched a DAO for the project (called a Squad) with the contributors to the project as members. Here, we will coordinate further development (such as building a user app or collaborating with artists, creators and art DAOs) and govern the future operations of the algorithm (such as value generation, ethical considerations etc.). With this proof of concept, we are now looking to kickstart more development by opening up a grants program for those with interesting ideas for AI x Web3 algorithms or apps.
https://miro.medium.com/max/1400/0*ruQXwsmc1PUpsAg-
We are interested in funding projects at the intersection of AI and Web3. The main criteria is that these algorithms or apps should have an innovative ownership model. They should create value for the contributors, as well as the users. While we want ideas to form from the bottom up by individuals, we give some ideas below to get the juices flowing. The easiest way to meet the criteria is to use a similar process and technology stack to the one that we have developed (it should be possible to use the hacking sessions and other resources as a guide). One example is to find an interesting private dataset on the Ocean marketplace and add some value:
There are many other decentralised data marketplaces with interesting datasets, such as dClimate and Energy Web, as well tools such as Dune Analytics.
Another idea is to build on top of existing machine learning algorithms and publish to the marketplace. For example:
Does this create more value compared to building on top of the original open source model in the traditional sense? Good question. We think it might open up interesting new possibilities. Excitement often builds around newly-released models (e.g. ArcaneGAN on HuggingFace recently). The networks that form have inherent value. What if we formalised these communities and gave its members the ability to capture this value e.g. using shared treasuries, governance and funding mechanisms?
Another idea is to build an app that integrates machine learning models with Web3 hangouts. We would personally love a channel in our Discord that converts text to image art using VQGAN+CLIP (just like EleutherAI) or other models. Or what about a voice channel with AI-generated music?
Perhaps you are interested in token engineering. So are we. Developing new economic models for (i) distributing the value generated by AI and for (ii) funding AI projects are huge problems that are not well explored. Where are the AI cooperatives that are owned by the workers? An example of a token engineering project proposal might be to:
If you have a commercial (rather than a technical) background, some ideas for proposals include: