Vint Cerf: Predicting the future as Robots and Humans
If robots could think like humans and act like humans, what would happen to the world as we know it? Would the world become a better place? Or would unforeseen problems arise? This may be tricky subject, but Vint Cerf is the one with answers.
Cerf, universally regarded as one of the “co-fathers” of the Internet, presented “What IF Machines Thought Like Humans?” as part of Purdue University’s Ideas Festival, the centerpiece of the University’s Giant Leaps Sesquicentennial Campaign. Cerf’s talk aligned with the festival’s theme of AI, Algorithms and Automation: Balancing Humanity and Technology. The April 5 event held was sponsored by the College of Engineering.
Comedian in a Three-Piece Suit
The hourlong event, hosted and moderated by Mung Chiang, Purdue’s John A. Edwardson Dean of the College of Engineering, was broken up into three sections, including an audience Q&A session at the end.
The IF event in Stewart Center’s Fowler Hall started with a roaring applause when Cerf, adorned in his signature three-piece suit, took the stage.
“You know when people clap before you say anything it’s tempting to sit down because it won’t get any better. So first of all, I’m not using slides at all, because my favorite expression is power corrupts and PowerPoint corrupts absolutely,” he said to a full house of laughter.
In a talk that was equally comedic, insightful and inspiring, Cerf described his life at Google and the work he does with machine learning. The presentation focused on his advocacy for knowing exactly what we are teaching our machines to think and how to be careful with the autonomy we give them.
The Life of Cerf
“Vint is widely regarded as a ‘Father of the internet,’ who co-authored with Bob Kahn, the 1974 landmark paper which described the foundation of the internet as we know it today — the TCP/IP protocol,” said Chiang.
Cerf, along with his development of the Transmission Control Protocol/Internet Protocol (TCP/IP), predicted how AI and technology will affect future societies environmentally, economically and socially. He started his renowned career for the U.S. Defense Advanced Research Projects Agency (DARPA), where he worked to create the TCP/IP technology.
When the concept of the internet was introduced during the 1980s, Cerf joined telecommunications giant MCI to help develop its commercial email system. During his impressive career, Cerf has earned numerous awards including the National Medal of Technology, the A.M. Turing Award, the Presidential Medal of Freedom, the Marconi Prize and the Japan Prize. Today, he serves as Vice President and Chief Internet Evangelist at Google.
“I would like to give a summary and explanation of my title at Google as Chief Internet Evangelist. Some people misunderstand that as a religious title and, in fact, when I respond to this statement, I tell them I’m Geek Orthodox,” Cerf said.
Human Thinking vs. Machine Thinking
“The topic, in my brief introduction here, is about artificial intelligence. I confess to you I always thought artificial intelligence might be best described as ‘Artificial Idiocy,’” Cerf said. “Machine learning and multi-layer neural networking are not necessarily the same as general artificial intelligence. General artificial intelligence has to do with the ability of a system to take a lot of input and formulate a real-world model of a conceptual idea in order to reason about the model.”
Cerf asked the audience to think of a simple table to further explain the difference between human thinking and a machine’s way of thinking.
He explained that we do not usually think of a table as a “flat surface that is perpendicular to the gravitational field” but that is what it is. Instead, we think of a table as an object to sit things upon. After humans understand the properties of this table, then they can easily identify other objects that can be used as tables. This is how humans can generalize abstract concepts of things quickly compared to a machine where abstract concepts are harder to learn from a simple input.
“As humans, we take-real world objects, abstract these models, reason about them and then apply what we have learned. I find that way of thinking to be missing in most artificial intelligence projects,” Cerf said.
Purdue industrial engineering student Simran Kachroo was captivated by Cerf and his views of machine learning. “I think it was really interesting and inspiring listening to Cerf talk about how humans think compared to machines that they program. The ‘intelligent machine’s that they create use data that is fed to them, yet they do not have the capabilities to really apply abstract concepts or go further than generalization. That was my favorite part of the event.”
Concern for Teaching Machines to be More Like Us
For much of his career, Cerf has been an advocate for safe programming, being conscious of what we put into multi-layer networks and cybersecurity. These issues have been important to him and his work because of how much society depends on reliable and safe technology.
“We should have concerns about how we apply these technologies and what we rely upon them for. I am not arguing that all machine learning is dangerous or bad, but I think we should be aware of its potential,” Cerf said.
Emily Petersen, who is pursuing a master’s degree in aeronautical and astronautical engineering, was pleased to hear that someone in the tech industry is being responsible for their robot programming. “I liked that Cerf has a strong sense of self-awareness about how he creates his technology and the restrictions that should be placed in machine learning.”