Alan Turing created the test that bears his name more than 70 years ago and which has become an obstacle that artificial intelligence must overcome in order to be considered truly intelligent: a human and a machine speak and if the person cannot tell if it’s an AI, so it passes the test.
However, the founder of DeepMind, Mustafa Suleyman, considers that this test could be outdated: “It is not at all clear if it is a significant milestone or not,” he said. Suleyman believes that “we care not only what a machine can say, but also what it can do.”
For this reason, in his next book, The Coming Wave: Technology, Power, and the Greatest Dilemma of the 21st Century, he proposes the “modern Turing test.” It consists of the following: you give an AI $100,000 and see if it can turn the initial investment into a million dollars. The machine must research an e-commerce business opportunity, generate blueprints for a product, find a manufacturer on a site like Alibaba, and then sell the item (with a written description) on Amazon.
When that happens, “the consequences for the world economy will be seismic,” Suleyman says, since the traditional Turing test “tells us nothing about what the system can do or understand, nothing about whether it has established complex internal monologues or can engage to plan in abstract time horizons, which is key to human intelligence.
Suleyman affirms that the technological industry focuses on the remote possibility of achieving artificial general intelligence (IAG), that is, algorithms with cognitive capacities that equal or exceed those of humans, something really complicated and outside the current possibilities.
In his opinion, the most achievable and significant short-term goal is what he calls capable artificial intelligence (ICA): programs capable of setting goals and performing complex tasks with minimal human intervention.
