RSS feed Get our RSS feed

News by Topic

BizReport : Internet : May 05, 2017

Expert: The dark side of AI for brands

Artificial Intelligence has been touted as a great opportunity for marketers to better reach and engage with consumers. But, according to one expert, there is a darker side to AI, and that side could harm the brands hoping for help.

by Kristina Knight

Kristina: While most of us are focused on the positives of AI - better targeting, more personalized experiences - there is a darker side to the use of AI. Could you explain more?

Michael Fauscette, Chief Research Officer, G2 Crowd: The simplest and one of the most dangerous dark sides relates to security and hacking. With the increasing growth of hacking / criminal activity including government sponsored efforts, the probability that AI will be used to create anything from adaptive or situational malware that learns and changes as you try to detect and fight it; to malware that invades a network or system and sits in the background learning the best time to deliver its payload. Intelligent security is the white hat version of AI and its opposite or black hat, adversarial machine learning will surely kick off an arms race.

Kristina: But the deep learning opportunities through AI are supposed to be a help.

Michael: The developments with deep learning are exploding with applications like image captioning or facial recognition to self driving vehicles. The problem though, freely admitted by the creators, is that as the machine learns we lose the ability to understand what it is doing to create its outcomes. They don't understand how the models work.

Kristina: Can't AI based image recognition classifications simplify the workload?

Michael: [AI based image recognition systems or classifiers] are being used for facial recognition, image captioning, self driving cars and a host of other things. The opposing dark side to these is that it appears that with the "right" image they can be fooled or manipulated. While still mostly experimental (I think anyway), researchers have managed to fool facial recognition systems among other things. The riskier applications though, like convincing a self driving car to drive into a tree by placing an image in its path, or even using the technology against a future AI enabled weapon are very concerning. This is another variation of the adversarial machine learning label.

More from G2Crowd and Michael next week, including how to use AI while mitigating these risk factors.

Tags: advertising, AI advertising, AI trends, artificial intelligence trends, G2 Crowd, internet trends

Subscribe to BizReport



Copyright © 1999- BizReport. All rights reserved.
Republication or redistribution of BizReport content is expressly prohibited without the prior written consent.
BizReport shall not be liable for any errors in the content, or for any actions taken in reliance thereon.