AI applications are being built by thousands of companies that cover the gamut of ways to use the software, something Intel showed off this week at its annual developer’s conference. Concepts from the highly abstract to the more tangible emerged on the showfloor at Intel Innovation, attended by thousands of developers in San Jose.
Here are two of those apps, picked to describe the dynamic range range of use cases. In one example, Intel is using AI to help the US DARPA (and potentially all kinds of security groups ) learn when their vision recognition software could fail, in hopes developers will build a fix. In the other, a young innovator programmed his computer to recognize when he was slouching at his computer.
The latter app, called “StraightenUp” uses a laptop’s camera to detect when a user’s posture is off. When it recognizes something out of alignment by the human sitting at the keyboard, the computer blurs the display. Straighten up, and the display gets sharp again.
Created by Maciej Jalocha, 19, from Poland, the app is based on a browser extension that he wrote and AI code he wrote and added on top. Jalocha, 19, entered his concept in a competition sponsored by Intel called the AI Global Impact Festival and was on hand in San Jose to accept an award, which included a certificate and a new laptop. He joined other young finalists with a range of other compelling ideas.
When I tried out StraightenUp, it lagged somewhat as it transitioned from blurry to clear, but it still worked. No way was I going to finish my work on that computer without straightening up, I thought. For me, it brought reminders of my teenage years as I would sit studying at my desk and my father would quietly enter my room and massage my shoulders into proper form. My mother was more direct, shouting, “Sit up straight!” when she’d see me slouch. Somehow, a computer reminder seems a little more subtle.
Jalocha, who is studying data science in Copenhagen, said he would like to continue work on StraightenUp and customize the application to allow users to set a sensitivity level.
The other extreme use of AI was shown by Intel Labs researchers, showing off ways engineers designed an adversial AI graphical pattern that could be used by bad guys to break visual recognition capabilities. On a large display, the graphical pattern caused the AI human recognition program to fail to identify a human, causing jitter on the screen.
Conceptually, someone could try to fool a visual scanner used at the entrance of a secure facility by donning a suit of clothes with the graphical pattern to confuse the AI software about what it saw, Intel said.
Intel is working with the DARPA on the concept. What’s most interesting about the application is that Intel used AI to create the distorting graphic. AI being used to fight AI, in other words.
AI Global Impact winners
Intel’s third annual AI Global Impact Festival attracted student entries from 26 countries who competed for more than $500,000 in prize money. At Intel Innovation, nine global award winners were named and honored. In addition to StraightenUp, other winners included a team from Costa Rica that used computer vision and AI to help protect native bees and a team from Malaysia to create a smart city disaster digital twin to help save lives. Winning projects were subject to an ethics audit by Intel’s Responsible AI team.