It is said that a tool is neither good nor bad. So a knife can be used to chop a carrot or kill a person. Nevertheless, with each new progress made in the field of artificial intelligence, a whole horde of screaming technophobes puts the outcry. Artificial intelligence is too reminiscent of the fictional Skynet and fear it must passionately. Or maybe not?
The last milestone reached by the University of Chicago proposes a scenario in which the people are unable to distinguish one human language generated by a neural network. We will grant the skeptics who certainly is a valid paradigm for a Hollywood blockbuster. But apart castles in the air, the reality is that the way in which it has been found that uniqueness is a Pandora box for the restaurant business.
Teaching a neural network to write fake reviews
It is well known that reviews applications They are a double-edged sword. On the one hand, serve to establish a online presence, inform the public quality of services provided and keep you ahead of competitors. But, on the other hand, They can be abused establishments to harm others or inflate own score undeservedly.
Yelp It is one of these apps. The best known in the US. THE. and one of the most important also in the hispanosfera. With this in mind, Chicago researchers set out to test whether the use of artificial intelligence It could be used to generate avalanche spammy undetectable eyes of users.
First, the code that would allow machine learning written. Yuanshun Yao, first author of the scientific study, received more than four million analysis Yelp and he introduced into the input nodes of innovative software. With a volume of such data, the neural network generated original reviews without difficulties.
To check if a person perceived as false opinions, analyzes were scored by a team outsourcing services through the famous market microtrabajos Amazon Mechanical Turk. The texts generated by artificial intelligence They were statistically indistinguishable from the real.
Although generally It is difficult to create compelling language using neural networks, the team led by professor in computer science Ben Y Zhao It has exceeded expectations by applying a three-step methodology: identifying relevant contextual information, the association of words related to the context and location of suitable replacement words.
The result is disturbing.
I've been going to this restaurant several years now and have never had a bad experience. The attention is fantastic! They are always friendly and attentive. I will definitely return, I think I'll be back!- said exemplary machine.
-¡NO LOSE YOUR TIME AND MONEY! Worst service ever experienced. This place is a joke. The waitress was rude and said he would call the boss, but that never happened. I wish I could put zero star- complains when forced to write a bad review.
¿Averiguarías are false if they were in the midst of so many?
Question remains whether this improvement in our ability to interpret text automated have beneficial use or bring new headaches. Is a must mention point out that, and we can write fake reviews, it is also possible use this technology to detect fake reviews. In any case we can let out a snort of relief because currently this technology is not available to everyone.
Buy positive ratings for handling online reputation
Buy Positive Feedback is definitely possible. Not only can buy these services for Yelp; Google Maps, TripAdvisor, Facebook and many others who are also in deep water. The deluge of false opinions is incessant.
It is logical that businesses risk being penalized when one point difference in valuation It means an increase or decrease (as appropriate) between five and nine percent in profits. For these premises it is important to box with alacrity as possible, the negative impacts They are really devastating.
Vivek Devrari, cofounder HotelXsell, recounts with satisfaction seeing a local restaurant down the top of the classification to a position slightly above the three-hundredth. The impact came when the MAC address, from which the reviews were published Rigged, It was detected. Vivek also adds how a automatic feedback system well designed have a most significant impact and longer, It is fully in accordance with the terms and conditions of use of the mentioned applications.
Brian Bergstrom, of the team Yelp Elite Squad, y Jay Skowron, specialist online reputation management, seconded this opinion. Yes, a simple Internet search returns a ghost writers thousand and willing to write false opinions for less than twenty euros. But it is also true that applications turn against this type of behavior and the result is usually a final purge.
Much more painful is to know the specific cases affected owners. We have known the situation Alberto, whose business is bombarded by false analysis They are leaving their establishment to the bitumen. You can not do anything about it and he resigns himself to remark sarcastically that “No opinions buy [TripAdvisor]”.
Complaints are multiple: catering establishments receiving reviews several months after closing, Restaurants in works allegedly being satisfied diners throughout the reform and more. Everything a circus! And dissatisfaction with these applications can only increase ...
We can only hope that the inevitable falsehoods campaigns led by artificial intelligence and dissemination of sites with credentials validated reviews BBB Customer Reviews arrive synchronously. Otherwise the arrival of AI can be a real catastrophe for these portals.