It is surprising that after almost 70 years of its creation, the famous Turing Test to evaluate the ability of machines to ‘think’ has not evolved into a social format integrated with the cloud environment, following the evolution of computers and the Internet itself.
I know that there are initiatives, such as the Loebner Prize, to bring the competition of various technologies and evaluate their quality through a specific selection, but in my opinion, and because of the relevance of Artificial Intelligence today — after a long winter — it’s necessary we think about an evolution for this test, so fundamental to evaluate the state of the art of our machines and technologies in terms of representing, understanding and communicate like us, humans.
In this sense, I would like to propose an evolution to the Turing Test, as described below, which I would call the Collective Turing Test.
The idea is simple: create an online/real-time platform, which can be a website, where there would be the main actors of the test, but acting in a collective and collaborative way, with a permanent ranking, evaluating the performance of all actors.
For example, ChatBot systems would be one of the actors, seeking to establish communication with people so that they actually ‘look’ human beings. Other actors would be humans looking for a test or discover good ChatBots. And other humans would be ordinary people, just in search of communication or make friends, as in any other chat system, without necessarily seeking to evaluate the machines.
The evaluation/rating systems would be both automatic, using Artificial Intelligence – obviously – or by the users, with a fast quality control survey after the chat.
In other words, a large machine for collective and collaborative testing of machines, forming a new ‘Imitation Game’, where, sooner or later, we would see the ranking of machines and systems that would really surprise everyone, with the ability to ‘think’ like us, humans.
By Rogerio Figurelli at 05/30/2020