AI has the power to learn, grow

AI may soon auto-generate music, books, movies and news. This has the potential to take fake news to the next level and create a scenario wherein citizens may not distinguish fact from fiction

IF WE ARE GOING TO CREATE AI THAT WORKS FOR ALL PEOPLE, WE NEED DIVERSE GROUP OF PEOPLE BUILDING IT — MELINDA GATES

MACHINES LEARN THE SAME BIASES OF THEIR CREATORS BECAUSE IT IS ONLY AS GOOD AS THE DATA IT LEARNS FROM

AN ONLINE BEAUTY CONTEST CALLED BEAUTY.AI, RUN BY YOUTH LABORATORIES WITH PARTNERS LIKE NVIDIA ON THE CONTEST WEBSITE CHOSE PREDOMINANTLY WHITE FACES AS WINNERS

GOOGLE’S PHOTO SOFTWARE INFAMOUSLY TAGGED TWO BLACK USERS ‘GORILLAS’ AS THE DATA LACKED ENOUGH EXAMPLES OF PEOPLE OF COLOUR IN 2015

MICROSOFT RELEASED ITS AI CHATBOT TAY TO LEARN HUMAN BEHAVIOUR BY INTERACTING WITH OTHER TWITTER USERS. WITHIN 16 HOURS, TAY WAS SHUT BECAUSE ITS TWEETS HAD BECOME SEXIST AND PRO-HITLER MESSAGES

AI ALGORITHMS SAID 68 PER CENT WOMEN ARE LIKELY TO APPEAR IN COOKING PHOTOS, WHILE A HUMAN EXAMINATION FOUND THAT 33 PER CENT PEOPLE COOKING ARE MORE LIKELY TO PICTURE WOMEN, AS PER PHD STUDENTS AT THE UNIVERSITY OF VIRGINIA AND UNIVERSITY OF WASHINGTON

COMPANIES ARE DOING ENOUGH TEACHING ROBOTS TO MIMIC HUMAN BEHAVIOUR, BUT NOT ENOUGH ABOUT TO TEACH WHAT IS APPROPRIATE BEHAVIOUR

GOOGLE TRANSLATES DISPLAYED SIGNS OF SEXISM BY SUGGESTING WORDS LIKE ‘HE’ FOR MALEDOMINATED WORKS AND VICE VERSA

INITIALLY THE CREATORS OF VIRTUAL ASSISTANTS THAT EXECUTE BASIC TASKS LIKE APPLE’S SIRI, AMAZON’S ALEXA ARE GIVEN FEMALE VOICES WHILE MORE SOPHISTICATED PROBLEM-SOLVING SYSTEMS SUCH AS IBM’S WATSON, MICROSOFT’S EINSTEIN WERE GIVEN MALE VOICES

WHAT ARE COMPANIES DOING TO ADDRESS BIAS FACEBOOK RECENTLY FORMED A SPECIAL ETHICS TEAM TO PREVENT BIAS IN ITS AI SOFTWARE GOOGLE’S DEEPMIND AI GROUP FORMED AN ETHICS AND SOCIETY TEAM TO ADDRESS THE ISSUE OF BIAS MICROSOFT’S RESEARCH ORGANISATION ESTABLISHED A FAIRNESS ACCOUNTABILITY TRANS-PARENCY AND ETHICS GROUP