Newsroom: Highlights

Bookmark and Share Email Print

03.18.19 - Vint Cerf visits CALIT2

Vint Cerf speaks to an audience at CALIT2



Much of the credit for the modern internet must go to Vinton Cerf. In 1974, Cerf, along with fellow engineer Bob Kahn, designed the first internetwork protocol – a means of communications that allowed different types of pre-internet networks to communicate with each other. They also came up with the name "Internet," which meant "the internetworking of networks."

On Monday, March 18, Cerf, spent the day at CALIT2. He met with the institute’s director, G.P. Li, toured CALIT2 labs, and delivered a presentation to students, faculty, staff and guests.

After touring the facility, Li commented that Cerf had offered “a lot of constructive feedback and direction about what we should work on next.”
 

(left) Cerf started the day at iGravi (Interactive Graphics and Visualization) Lab where he previewed professor Aditi Majumder’s immersive virtual reality environments. One of her projects uses multiple projectors that illuminate large surfaces to create a seamless virtual environment in which one or more users can navigate and interact.
(Right) At the eHealth Collaboratory, Cerf tried out FlintRehab’s MusicGlove with the assistance of student, Sharon Dinh. The hand therapy device is designed to help with mobility rehabilitation. MusicGlove works by motivating users to perform hundreds of therapeutic hand and finger exercises while playing an engaging musical game.

Cerf concluded his visit by speaking to an audience in the CALIT2 auditorium. The topic of his presentation: “Augmenting Human Intellect: Prospects.”

“Augmenting Human Intellect” is a concept that’s been around since the beginning of the Internet.

In 1962, Douglas Carl Engelbart – best known for inventing the computer mouse – published "Augmenting Human Intellect: A Conceptual Framework.” The paper outlined his vision of the future as "increasing the capability of a man to approach a complex problem situation, to gain comprehension to suit his particular needs, and to derive solutions to problems.”

Cerf believes this vision still exists today – “to create, understand, invent and do. People are tremendously excited about what they can do,” he said.

But there is immense difficulty in mimicking human intellect, even for casual AI use, he said. As an example, Cerf recalled an early experience in developing a language translation application. It seemed like a simple enough problem. Connecting users to a Russian dictionary database should provide an instant and useful translation of English to Russian. In initial testing, the group input the phrase “out of sight, out of mind.” The application promptly supplied the translation: “invisible idiot.”

Cerf viewed a demonstration of current AI research projects in
CALIT2’s Data Engineering ThinkTank.

Regardless of difficulties, he predicts fascinating advancements in AI tools.

One needs only to observe a Google search to see how machine learning is advancing, he said. Rather than typing a search term into Google, he said, “I started asking questions in the search field and got better results. We’re using all of these tools without giving it much thought.”

Breakthroughs in neural interfaces have already offered marked improvements for health-aid devices like cochlear implants.

Cerf predicts even more fascinating advancements in AI tools. Sensory-neural interfaces will allow the development of ocular implants that can restore sight for many types of blindness. Sensory-motor interfaces will bring about artificial limbs and hands, even spinal injury bypass devices. “Some people are hoping for a memory implant, but we haven’t gotten there yet,” he added.

Cerf also sees a future filled with challenges. AI has risk factors and limitations, such as bias, disinformation and botnets. Some believe the proliferation of spam and cybercrime is, in part, the failure of early internet pioneers to predict what could happen as a result of the internet’s open infrastructure. Moving forward, Cerf urges AI engineers to employ a standard of ethics, and commit to developments that are transparent, explainable, reversible, trainable and human-led. “All decisions should begin and end with human decision points,” he concluded.

-  Sharon Henry