It seems that now we are not going to be able to trust the product reviews that we find in online product and service stores. Since researchers from the University of Chicago have managed to develop, thanks to the artificial intelligence (AI)a software that makes it possible to write automatically fake reviews that result extremely believable.
Specifically, this new application of technology represents a major threat for online sites like or , where user ratings are crucial and a proliferation of fake reviews could affect your credibility.
Artificial intelligence capable of giving an opinion as if it were a real person
The has been carried out by a group of researchers from the University of Chicago and aims to demonstrate how an algorithm developed with AI and the technique of deep learning (RNN or recurrent neural networks) can be used to create undetectable fake reviews.
After asking to 600 Internet users On several online reviews on Yelp, they found that users not only did not think that fake reviews were real, but also came to consider them tools.
And what is more worrying, the fake reviews generated by the algorithm were also not identified by programs. specialized in plagiarism detection. The researchers note that this demonstrates that the algorithm can acquire the ability to opine as if it were a real client and influence other users.
How is a fake AI review like a real review?
An example of false review could be this: “I love this place. I went with my brother and we ordered 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 little place with a great deal.”
At first glance, we can see that there is nothing strange about this opinion. First he tells his experience and then makes a couple of recommendations so that people go to visit the site. Although the last sentence is a bit strange, it doesn’t differ too much of an expression you can use to use a any user after visiting the place.
Fake reviews: a threat to the credibility of the Internet
It is true that currently, products and services are already spread, but the cost in time and money invested is much higher than that which would be produced with software of these characteristics. This means that in the case of normalizing these practices, their massive spread over the Internet.
In any case, the researchers point out that despite everything, there may still be a way to detect the fake reviews. While the opinions created by the software may appear identical to those of a real person, there are still some subtle differences – such as the separation of characters or letters – that a properly developed program could detect.
A technology that could be used to create other malicious content on the Internet
Ben Y Zhao, a professor of computer science and one of those responsible for this study, has stated that he is not aware of any cases in which AI has been used to generate malicious falsifications of opinions. However, he acknowledges that a competent person on this subject it would not cost you too much to develop a similar system to your research.
On the other hand, Zhao stresses that the most worrying thing is not that an algorithm can write false reviews on Yelp or Amazon. He warns that this technology can be used for other purposes, for instance “entire articles written on a blog by a robot”. In that case, you would have to “Analyze where the information comes from, how it can be verified, etc.”
The study is titled “Automated Crowdturfing Attacks and Defenses in Online Review Systems” and will be presented at ACM Conference on Computer and Communications Security later this year. If you want to know more, you can read the full report .
Image:
Stay informed of the most relevant news on our news channel
Facebook Comments
Disqus