Artificial intelligence is changing our lives and algorithms are changing the way we communicate, but humans are still needed in the loop because the data that powers AI systems can have biases.
That was the message from Nicholas Thompson, editor-in-chief of Wired magazine, who shared a cautionary tale of how Mexican restaurants were receiving dramatically lower ratings than other eateries in their areas. Turns out, the artificial intelligence behind the website had been programmed to recognize the word “Mexican” frequently was used in a negative context, and therefore the AI concluded a Mexican restaurant was bad in some way.
“Do we trust machines to make decisions? The data that powers AI systems can have biases,” Thompson said Tuesday. “Most of the world is focused on artificial intelligence. We need to have a national conversation in the United States about the use of data, but we are not doing it.”
According to Thompson, technology is reshaping society “in every way.” He noted people who are infertile soon may soon able to have babies by creating their own stem cells from any adult cell. Similarly, if a person has died but left DNA on a hairbrush, someone still might produce a child with the deceased.
“This is coming, and we will have debates,” he said. “One couple met on virtual reality, fell in love and got married. Wired covered the wedding, which had guests attending in virtual reality. Science doubles every 18 months, same as computing capacity.”
Thompson warned data privacy is a topic that had received far too little scrutiny. He said there was “15 years of under-reaction to lack of privacy, followed by a month of overreaction” following the Cambridge Analytica/Facebook scandal connected to the 2016 presidential election.
Algorithms are changing the way humans communicate, Thompson continued. He said the 2016 election showed how Facebook created algorithms to deliver news based on anger. More recently, Instagram created algorithms to remove mean comments and/or emojis. The AI was programmed to identify comments that were mean, racist or sexist. In the same vein, Google is building programs to identify terrorist content.
“Think about the implications for free speech,” Thompson said. “This is a country founded on the philosophy of free speech. But instead of the Supreme Court deciding the limits of speech, it is Instagram.”
What about jobs in the future given the transformation of AI? Thompson noted robots and computers are surpassing humans in many ways, and the number of tasks they succeed at is increasing. “It will get to the point where machines will be better than everything,” he said. “But it will not all be bad. ATMs were supposed to mean the end of tellers when they were introduced in the 1970s, but they improved banking. There are more tellers today than there were in the 1970s, and because they are not dispensing cash they have more interesting tasks to do. We have worried about machines taking jobs from humans for hundreds of years.”
According to Thompson, there are choices people need to make now that will have huge consequences later. “AI will do incredible things, but we need to make moral and sophisticated choices. There is never a moment you catch up with technology, it is always changing.”