Prof. Irad Ben-Gal from the Department of Industrial Engineering, the head of The Laboratory for AI, Machine Learning, Business & Data Analytics (LAMBDA) was recently interviewed by Walla News on the risks of Deep Fake technologies during the elections. Prof. Ben-Gal explains that application of Deep Fake technologies can be legitimate and have a positive effect if properly used. Among other things, he discussed deep network models and the options nowadays to create accurate simulations of reality in applications such as smart cities and epidemiological models of infection.

The study

Conducted in collaboration with Dr. Erez Shmueli, from the Department of Industrial Engineering, this study showed the most effective ways of disseminating information was by using regular people in your social network who respond and discuss new opinions that you have just been exposed to. Opposite to celebrities, when regular users retell and circulate the new information, it creates local effect that makes the information more apparent on the web. This was the tactic that Cambridge Analytica used to target Facebook users during the first campaign of Trump.

This study was done as a follow up to the research of Dr. Alon Sela, who holds PhD from the Industrial Engineering Department at Tel Aviv University in collaboration with Prof. Shlomo Havlin and Dr. Louis Shechtman. They examined models of spreading opinions in modern society, and in their research, they compared two spreading mechanisms: “Word-of-Mouth” (WOM) and online search engines (WEB). In their study, the researchers applied modelling and real experimental results. This way, they compared opinions that people adopt through their friends and opinions they embrace when using a search engine based on the PageRank (and alike) algorithm.

The research findings showed that the opinions adopted through the use of WEB scheme is more homogenous. People using the search engines follow just a few dominant views without having a variety of different opinions, a phenomenon known as “the richer gets it all”. In contrast, WOM scheme produces more diverse opinions and their distribution.

Combining the physical world with the digital world

​The “Digital Living 2030” project, spear-headed by Ben Gal and Professor Bambos at Stanford University, is concerned with connecting the digital world and the physical worlds as trends are analyzed about our future lives 10-15 years from now. Ben Gal explains that an avatar (digital character) can be created to represent a person in the digital space for a positive purpose. “Imagine an expert radiologist whose digital character has learned from all his past data about medical imaging so that others can work using his digital character when his physical character is no longer present, for example, when he sleeps at night.”

The research helps better understand the situation with the Deep Fake technology. Since the technology is available now for commercial use, it can be “used and abused” by anyone exposed to it. Fake videos can be produced in big amounts by ordinary people that are connected on social network and can affect each other`s opinion.

In the last days of an election campaign, this kind of activity can dramatically change the results, and “even if the fraud is detected at the end of the day, it will be too late.” – says Prof. Ben-Gal. “From the other side, having a diversity of opinions, new and not ordinary, is essential and enriching.  If the application of technology is legitimate and reliable, it can enrich the body of human knowledge with diverse opinions.”

Loving the problem is the greatest way to invent

The interdisciplinary study was led by Idan Fishel, a joint master student under the joint supervision of Dr. Ben M. Maoz of the Iby and Aladar Fleischman Faculty of Engineering and the Sagol School of Neuroscience, Prof. Yossi Yovel and Prof. Amir Ayali, experts from the School of Zoology and the Sagol School of Neuroscience together with –, Dr. Anton Sheinin, Idan, Yoni Amit, and Neta Shavil. The results of the study were published in the prestigious journal Sensors.

The researchers explain that at the beginning of the study, they sought to examine how the advantages of biological systems could be integrated into technological systems, and how the senses of dead locust could be used as sensors for a robot. “We chose the sense of hearing, because it can be easily compared to existing technologies, in contrast to the sense of smell, for example, where the challenge is much greater,” says Dr. Maoz. “Our task was to replace the robot's electronic microphone with a dead insect's ear, use the ear’s ability to detect the electrical signals from the environment, in this case vibrations in the air, and, using a special chip, convert the insect input to that of the robot.”

To carry out this unique and unconventional task, the interdisciplinary team (Maoz, Yovel and Ayali) faced number of challenged. In the first stage the researchers built a robot capable of responding to signals it receives from the environment. Then, in a multidisciplinary collaboration, the researchers were able to isolate and characterize the dead locust ear and keep it alive, that is, functional, long enough to successfully connect it to the robot. In the final stage, the researchers succeeded in finding a way to pick up the signals received by the locust’s ear in a way that could be used by the robot. At the end of the process, the robot was able to “hear” the sounds and respond accordingly.

“Prof. Ayali’s laboratory has extensive experience working with locusts, and they have developed the skills to isolate and characterize the ear,” explains Dr. Maoz. “Prof. Yovel's laboratory built the robot and developed code that enables the robot to respond to electrical auditory signals. And my laboratory has developed a special device - Ear-on-a-Chip - that allows the ear to be kept alive throughout the experiment by supplying oxygen and food to the organ, while allowing the electrical signals to be taken out of the locust’s ear and amplified and transmitted to the robot.

“In general, biological systems have a huge advantage over technological systems - both in terms of sensitivity and in terms of energy consumption. This initiative of Tel Aviv University researchers opens the door to sensory integrations between robots and insects - and may make much more cumbersome and expensive developments in the field of robotics redundant.

“It should be understood that biological systems expend negligible energy compared to electronic systems. They are miniature, and therefore also extremely economical and efficient. For the sake of comparison, a laptop consumes about 100 watts per hour, while the human brain consumes about 20 watts a day. Nature is much more advanced than we are, so we should use it. The principle we have demonstrated can be used and applied to other senses, such as smell, sight and touch. For example, some animals have amazing abilities to detect explosives or drugs; the creation of a robot with a biological nose could help us preserve human life and identify criminals in a way that is not possible today. Some animals know how to detect diseases. Others can sense earthquakes. The sky is the limit.”