An 11 page research paper was released earlier this month that details how to use AI as a means of detecting ICO frauds and scams.
The whitepaper, titled “IcoRating: A Deep-Learning System for Scam ICO Identification”, was put together by researchers from Stanford, University of California Santa Barbara, the University of Michigan, and leaad by Chinese based startup shannon.ai.
Identifying scams is considered by many to be very hard work with an incredible amount of research involved into ideas, team members, wording, and several different factors.
Scammers realize the work involved and gamble on the fact people won’t do their research properly and are often proven to be correct in this space.
The researchers believe they can use artificial intelligent machine languages to spot a fraud ICO pretty much at the click of a button.
A few members of this team spent months combing through white papers, code repositories on GitHub, cryptocurrency project websites, and biographical information of more than 2,000 ICOs. They determined whether the project was a scam and then labeled it as such using a computer program.http://www.cryptocurrency-truth.com/wp-admin/edit-comments.php
One of the project team members compared it to automating the task of determining spam email.
The team also examined how ICO’s performed after the launch to help with its analysis.
After adding the information to computer program it pulled out data points to find patterns which is how the program is essentially trained to pick up common signs of a scam.
According to one of the authors of the white paper report William Wang,
Some of the technology is based on established machine learning principles while other parts, such as assessing the backgrounds of a project’s founders, are more novel and specific to ICOs.
In the future, the project hopes to move away from making binary categorization and onto informing ICO investors what level of risk they may be taking.
The question however, really is, will such technology be used by organizations such as the SEC to automate the task of what they deem to be legal ICO’s, fraudulent ICO’s, and unregistered securities.
According to the whitepapers:
Compared against human-designed rating systems, ICORATING has two key advantages. (1) Objectivity: a machine learning model involves less prior knowledge about the world, instead learning the causality from the data, in contrast to humandesigned systems that require massive involvement of human experts, who inevitably introduce biases. (2) Difficulty of manipulation by unscrupulous actors: the credit rating result is output from a machine learning model through black-box training. This process requires minor human involvement and intervention.
If such a project becomes successful it will greatly help with informing users because no longer will one be able to claim that scam reporting is just “FUD” or an “attack by haters” but rather backed by actual computer predictions with no emotional involvement at all.