The Imitation Game: Startup's Big Idea in Building AI That's as Good as Human

gepubliceerd op by Cointele | gepubliceerd op

The premise is simple: an interrogator asks questions to a machine and to a human while sitting in a separate location, and receive the answers written down.

Now, a startup is taking on the challenge of building an AI which can succeed where no machine has done before - and is using blockchain to incentivize the public to help them in its quest.

A helping human handAccording to CEN, current AI offerings can be fragile - "Making childish errors when they come across new information from outside their training set." This is normally the key reason AIs struggle in the Turing test, as they often provide answers that lack any relevance to the question that was asked.

The startup's approach is to switch questions that an AI cannot answer to a human in real time - a seamless blend of machine and manpower.

Tokenization is used to log the answers given by "Minders," the humans intervening, and they are rewarded for their efforts in ERC-721 tokens - a non-fungible token that has become popular in recent years because of how they are unique and non-divisible.

Ownership of knowledgeThe startup firmly believes that AI will have a fundamental role in the 21st century, but argues that many machines are being built incorrectly - and subsequently lack human understanding.

Its platform also paves the way for individuals and businesses who are seeking expert information to pay for further insight from a human using tokens.

The company has developed a number of pre-production systems which aim to reveal the full potential of a hybrid artificial and human intelligence system.

Over time, CEN hopes to scale these platforms further - and ensure that humans who want to make a contribution can do so through an interface that's suitable for non-programmers.

It is hoped that artificial intelligence will help to reduce costs per hour when compared with the fees charged by humans alone.

x