Interview with Effect AI
We had the opportunity to meet and interview part of the core team of Effect AI, Jesse and Laurens, in the last NEO Devcon held in San Francisco.
Jesse introduced himself as the blockchain developer for Effect AI. While Laurens does the backend job. Both have studied computer science in Amsterdam. And have years of experience doing smart contracts since 2014.
Having worked together at Itsavirus, Jesse and Laurens were one of the first people to design an algorithm for automatic A/B testing, an AI driven technology used to automatically optimize business websites. They called this algorithm, The Effect. This is where the concept of Effect AI started.
It has come to their awareness the problems small AI company face. With the lack of computational power and a lot of data training involve, it is easy for private companies such as Google to swiftly monopolize the AI industry. In response to this, Amazon have devised a solution called the Amazon Mechanical Turk (M Turk), a platform where AI developers can source out human workforce to perform simple tasks in exchange of payments. This was indeed a great idea, except the 40% fee taken from requestors and a big delay in payment for the workers.
In order to break this oligopoly in AI, the team conceptualized Effect AI, a decentralized, open source platform for development of high-level artificial intelligence. The idea to tackle this problem is far from easy and most definitely a good infrastructure is needed. Effect AI has been compared to its numerous competitors but their main advantage, as Jesse and Laurens have both pointed out, is having a vertical structure tackling the problems in AI in 3 phases.
Phase 1 is their version of Mechanical turk derived from Amazon’s M turk without the soaring fees (yes its 0% fee) and delay in worker’s payment. But as mentioned earlier, Effect AI is not alone on this idea. Their phase 1 has been compared to Gems, another decentralized Mechanical turk under Ethereum platform.
This phase requires a lot of microtransactions in order to pay workers.
But also, Jesse has also highlighted Effect AI’s broader infrastructure and total package. “A prediction which has been made—which I think is true. In about 5 years, the mechanical turk market will be obsolete because AI is getting better in a fast rate and all these tasks will be performed by algorithms in the near future”. With that being said, Effect AI’s structure foresees this possibility as they have a place for algorithms in their project.
The phase 2 Smart market facilitates transition work from humans to algorithms. The problem AI developers are currently facing is the fact that AIs being created are only good at doing specific tasks. The Effect Smart market offers an opportunity for individuals or companies to sell and exchange their AI data. This open space allows anyone to build AI through the Effect AI standard. By providing easy access, interactions and exchanging of data with each other can quickly improve AI development and eventually can result in one emergent AI.
When asked about their comparison against Singularity marketplace. Jesse was quick to say that their marketplace unlike singularity, which focuses on a bigger mission, is more about practicality. Their aim and focus is more about step-by-step, creation of a market that resources algorithms that people can use and purchase. These algorithms would have linked wallets that the developers have access to. Every time an algorithm is used, it would require payment to their linked wallets.
Lastly, Phase 3 of their project is the Effect Power. This will be a distributed computational platform that will run deep learning frameworks with the likes of Deep brain chain, Caffe, and Golem. Although this phase is much further away from the current development. Jess and Laurens emphasized on the importance of using a valuable AI token which will be used from the M turk to the smart market and eventually will help run the computational power of Effect AI.
The team’s idea is to create a distributed, affordable AI platform that provides visibility to meet the demand in the AI industry. This is reflected by the token metrics decided by the team. At only $18 M hard cap, Laurens said that this choice was not based on the market value but what they need for the project to start running. They believe in equal distribution of token and hence, no private presale will occur and a personal cap will be implemented during the token crowdsale. Tokens will be immediately released for crowdsale but they have implemented a 2-year lock up period for the team and advisors of the project.
Both Jesse and Laurens showed genuine enthusiasm in tech and the future of AI and blockchain. Despite the hype in the project and the space, the team remains humble and focused on giving a fair and practical solution for AI development. Kudos and Good luck to the whole team!