Many of us have a love/hate relationship with restaurant review sites: they’re an incredibly useful resource for those of us who love to go out to eat, but at the same time, there are trust issues. How do we tell which are fake, bought or biased?
Those trust issues just got a whole lot deeper with the news that scientists in the US have trained AI bots to write fake restaurant reviews indistinguishable from the real thing.
According to Business Insider, researchers at the University of Chicago have trained a neural network to write, using thousands of real online reviews. The synthetic reviews virtually invisible to plagiarism software and were found to be "effectively indistinguishable" from real reviews in human trials. Here are three, all for the same restaurant in New York:
"I love this place. I went with my brother and we had the vegetarian pasta and it was delicious. The beer was good and the service was amazing. I would definitely recommend this place to anyone looking for a great place to go for a great breakfast and a small spot with a great deal."
“I had the grilled veggie burger with fries!!!! Ohhhh and taste. Omgggg! Very flavorful! It was so delicious that I didn’t spell it!!”
“My family and I are huge fans of this place. The staff is super nice and the food is great. The chicken is very good and the garlic sauce is perfect. Ice cream topped with fruit is delicious too. Highly recommended!”
Convincing, aren't they? The technology isn’t in the public sphere yet of course and the researchers were able to find ways to identify the AI generated text through, for example, the frequency of use of certain letters, but as it evolves, trust is going to be increasingly hard fought for.