Lovelace 2.0, the New Rival to the Turing Test
Ashley Allen / 2 years ago
The Turing Test – named for mathematician and computer programmer Alan Turing – is the world’s most famous method of assessing whether an Artificial Intelligence (AI) is capable of passing for human in conversation. Enter Lovelace 2.0, a new test developed by the Georgia Institute of Technology to test the creativity of an AI.
The Turing Test involves a panel of examiners engaging in a text conversation with both a human and a computer, not knowing which is which (known as a blind experiment). The assessors then have to guess which conversation was with a person and which with a computer. If 30% of the assessors mistake an AI for human, then it passes the Turing Test. Lovelace 2.0, however, aims to evaluate an AI’s capacity to produce art, in the form of poetry, a story, or a painting. The test takes its name from Ada Lovelace, the first computer programmer.
Professor Mark Riedl of the Georgia Institute of Technology explains, “For the test, the artificial agent passes if it develops a creative artefact from a subset of artistic genres deemed to require human-level intelligence and the artefact meets certain creative constraints given by a human evaluator.” He followed, “Creativity is not unique to human intelligence, but it is one of the hallmarks of human intelligence.”
Though Professor Riedl made the proposal in his paper, The Lovelace 2.0 Test of Artificial Creativity and Intelligence, the test is yet to be put into practice.