CHINA TOPIX

05/12/2024 03:44:27 pm

Make CT Your Homepage

New AI Test Challenges Turing Test

Lovelace 2.0

(Photo : Prometheus) The new AI test called Lovelace 2.0 challenges the Turing test.

An American professor proposes a new method of testing if artificial intelligence (AI) can have the same capabilities as humans.

To date, scientists use the Turing Test, named after famed computer scientist and pioneer Alan Turing, to evaluate if an AI can convince human judges it's human during a conversation.

Like Us on Facebook

Mark Riedl from the Georgia Institute of Technology suggests a new type of test for AI. The test will asks the robot to create a poignant poem, story or a painting. This new test is called "Lovelace 2.0", which is a new version of the previous Lovelace Test in 2001.

The test is named after Ada Lovelace, an English mathematician who wrote the first algorithm that was carried out by a machine. This achievement made her the first computer programmer in history in the 1880s.

The original Lovelace Test requires the AI to create something that it can't explain how it was created.

Riedl explains that in order for the AI to pass the new version of the test, it should create or develop a creative artifact taken from a subset of artistic genres and styles that utilize human like intelligence.

The creative artifact made by the robot should meet certain criteria given by the human curator. This could be a painting, a fictional story or an architectural design.

Although many algorithms have already created riveting stories and stunning paintings, no existing story generation system can pass the Lovelace 2.0 test, according to Riedl.

Experts in the field have mixed feelings about how the test might be made fool proof. Alan Woodward from the University of Surrey said this new test indicates creativity is a unique feature that makes humans, human and this separates humans from robots, at least for now.

A robot can only pass the famous Turing Test if it's mistaken for a human during a series of five minute chat conversations.

In June, a computer program supposedly passed the Turing Test. This program simulated a 13 year-old boy called Eugene Goostman from the Ukraine.

This chatterbot was developed in 2001 by a team from St. Petersburg, Russia but experts are still critical about the results of the test.

Real Time Analytics