Despite Facebook founder Mark Zuckerberg's repeated assurances that artificial intelligence will eliminate the proliferation of hate speech on the social networking platform, Mike Schroepfer, the company's chief technology officer and a formerly staunch advocate for AI, is no longer quite as sure.
In recent interviews with The New York Times, Mr. Schroepfer reluctantly admitted that the machine learning technology he once championed may not actually solve all of Facebook's problems. One of the main issues he described is that hate speech and other dangerous material come in so many forms, are constantly changing and are somewhat subjective, making it nearly impossible to train an algorithm to detect them — representative of the larger challenge plaguing the implementation of AI across all industries.
A prime example of this is the livestreamed video of the March shootings in Christchurch, New Zealand, which stayed on Facebook for several hours before it was finally removed. That video, Mr. Schroepfer said, did not resemble any of the content that Facebook's algorithms had been trained to flag.
Still, as newly flagged content like that video continues to teach the technology about new types of posts to remove, Mr. Schroepfer remains hopeful that AI can make a real difference. "It won't be fixed tomorrow. But I do not want to have this conversation again six months from now," he said. "We can do a much, much better job of catching this."
More articles about AI:
Dr. Eric Topol: Integrating AI into medicine will 'restore the human mind'
Facebook releases disease prevention maps to fight flu, malaria outbreaks
AI coach improves outcomes for teens in weight loss programs