Artificial intelligence? Only an idiot would think that
Web Log: Google’s hate-speech detecting algorithm, Perspective, is too easily fooled to be real AI
Prof Ian Bogost of the Georgia Institute of Technology: not every technological innovation merits being called “AI”
Not every technological innovation is artificial intelligence and labelling it as such is making the term “AI” virtually meaningless, says Ian Bogost, a professor of interactive computing at the Georgia Institute of Technology in the US. Bogost gives the example of Google’s latest algorithm, Perspective, which is designed to detect hate speech. While media coverage has been hailing this as an AI wonder, it turns out that simple typos can fool the system and allow abusive, harassing, and toxic comments to slip through easily enough.
Researchers from the University of Washington, Seattle, put the algorithm through its paces by testing the phrase “Anyone who voted for Trump is a moron”, which scored 79 per cent on the toxicity scale. Meanwhile, “Anyone who voted for Trump is a mo.ron” scored a tame 13 per cent. If you can easily game Artificial Intelligence, was it really intelligent in the first place?