Richard DeVaul Shares His Thoughts On Deepfakes
When asked about the phenomenon of deepfakes, Richard DeVaul said they were one of the most disruptive technologies in the modern world and not able to be dealt with at this current time. He further points out that this worries more than just himself and uses the Bulletin of Atomic Scientists moving the Doomsday Clock closer to midnight as an example.
The definition of “deepfake” is any media altered to represent someone else, commonly used to spread misinformation or for malicious intent. The origins trace back to 2011 during his tenure at Google X when Project Brain revolutionized the concept of machine learning. This same technology now has access to virtually every service in the world, from facial recognition to Alexa.
One of the learning systems at the core of deepfakes is GANs, fully known as the Generative Adversarial Networks. The paired entity works as one network trying to create a convincing fake while the other network learns how to distinguish the fake from the real thing.
This is the root cause of the rapid advancement of deepfakes, according to Richard DeVaul. He further mentions that because detector systems are improving, it ultimately causes the fakes to become better at seeming real.
Citing a documentary director’s use of deepfake technology to create audio that Anthony Bourdain never said as an example, DeVaul points out that the scenario could be a sign of things to come. He goes on to say that if evidence can so easily be called into question, it makes one wonder how decisions will be made in the future. He believes supporting journalism and facts is a start to countering the damage deepfakes could do in the future.
Richard DeVaul is a research scientist with years of experience and a former director of engineering for Google X.
His mission is centered on helping people understand the innovation culture. Richard seeks to use innovation as a tool to develop a real value for companies and the entire community.
In an article entitled “Want Innovation? Forget Invention, Learn To Execute”, Richard DeVaul explains Overvaluing invention and undervaluing execution leads to poor resource allocation and underperformance in most innovation-focused organizations. High-impact innovation comes from putting the pieces together well — design — and delivering a great product — execution. Go to this page for additional information.