AI that’s all well and good
ASU researchers present artificial intelligence initiatives that benefit society at national event
A woman works on a laptop displaying a prototype of a new chatbot powered by artificial intelligence, or AI. The chatbot is designed to help aging Americans experiencing memory loss. The work was developed by ‘YZ’ Yezhou Yang, an associate professor of computer science and engineering in the School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering at Arizona State University. It was one of two ASU projects presented at the inaugural meeting of the National Artificial Intelligence Research Resource Pilot, or NAIRR Pilot. Photo courtesy of Marcus Aurelius/Pexels
Ready or not, the era of artificial intelligence, or AI, has arrived. The scientific community is engaged in important discussions that are designed to answer one central question.
How can we make sure this emerging technology is used for social good?
Researchers at Arizona State University are playing a key role in providing an answer. Last month, two ASU faculty members traveled to the inaugural meeting of the National Artificial Intelligence Research Resource Pilot, or NAIRR Pilot. NAIRR is an initiative of the U.S. National Science Foundation, or NSF, to accelerate the discovery and innovation in AI to solve critical societal and global challenges. The effort supports more than 250 projects in 40 states.
Held in Arlington, Virginia, the NAIRR Pilot meeting brought together approximately 400 national AI innovators across industry and academia for poster presentations, demonstrations and expert panels.
‘YZ’ Yezhou Yang and Mina Johnson-Glenberg attended the invite-only event on behalf of ASU. Yang is an associate professor of computer science and engineering in the School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering at ASU, while Johnson- Glenberg is an associate research professor in the ASU Department of Psychology.
Both faculty members discussed impactful research underway at the university to create AI that provides real benefits to real people and encourages scientific advancement in AI.
Yang speaks on stage at the NAIRR Pilot meeting. He was one of two ASU researchers invited to present projects that demonstrate how AI can be used for social good. Image courtesy of Mina Johnson-Glenberg/ASU.
Down memory lane
Yang is a thought leader in computer vision, a type of AI where machines use sensors, cameras and complex software programs to perceive the world around them. As head of the Active Perception Group, or APG, a laboratory in the Fulton Schools, he directs research in autonomous vehicles, cognitive robotics and more.
He also studies how to use generative AI in innovative ways.
Generative AI, including popular tools like ChatGPT and Midjourney, refers to computer systems that learn from large amounts data in images, videos and text, then use instructions from users to generate new content.
At the NAIRR Pilot meeting, Yang presented his team’s work developing recall tools for aging Americans. For the project, the APG team created an AI-powered tool for those experiencing memory loss. The chatbot engages the user in personalized, soothing conversation and can generate helpful images.
The process is designed to mimic that of health care professionals who use images and visual cues when caring for patients with memory impairments, including those diagnosed with Alzheimer’s disease. Showing pictures to patients can promote a sense of calm and reduce feelings of social isolation.
“We believe AI can be a powerful resource to assist with memory recall,” Yang says. “Important future applications include helping aging Americans and their families with memory care.”
At the meeting, Yang sought to engage with other researchers on future developments for the project, showcasing ways to integrate user-provided media to make the tool more personalized and to incorporate video.
Mina Johnson-Glenberg, right, an associate research professor in the ASU Department of Psychology, presents her work at a poster session at the NAIRR Pilot meeting. Johnson-Glenberg’s team has created a new AI-powered physics tutor for mobile devices that provides personalized, Socratic feedback during motion graphing tasks. Photo courtesy of Mina Johnson-Glenberg/ASU
Putting AI-powered tutors in motion
Meanwhile, Johnson-Glenberg presented her own team’s project: an AI-powered physics tutor for mobile phones.
Johnson-Glenberg helms the Embodied Games Lab where her research team studies how to use emerging technologies, including augmented and virtual reality, electronic games and AI to improve learning and educational outcomes. She has been awarded seven grants from the NSF, some of which featured an AI mobile application that taught students to eliminate mosquito breeding in areas near their homes and schools.
Along with a team of three computer science students, Johnson developed a mobile application designed that helps students learn about motion in physics and how to graph the motion data they collect.
The project builds on the team’s prior creation of Motion Visualizer, a digital tutor that used light detection and ranging, or LiDAR, technology to track how students walk. The free mobile app uses data it collects from device sensors to encourage a student’s learning by interacting with their physical surroundings. This active way of learning, known as embodied learning, helps increase graph comprehension.
But the new project underway using NAIRR funds contains important advancements to the earlier technology. First, it uses multimodal data, or different kinds of feedback from the student’s phone. For example, when explaining the concept of position versus time, the app guides the student through an exercise where they move their body and their phone, collecting speed and movement data to generate a real-time graph.
It also incorporates a new form of Socratic AI. Named for the ancient Greek philosopher, the Socratic method uses open-ended question and answer sessions to help students learn through dialogue. Here, learners can ask questions on their phone. The AI tutor guides a discovery process, helping users find answers on their own through reasoning.
Johnson-Glenberg says it has been harder to train and fine-tune the application’s model than the team originally thought.
“The system usually wants to give student the explicit correct answer,” she says. “It is a dance to get the student to deduce what is wrong.”
Yang and Johnson hope that their participation in the NAIRR Pilot meeting will open doors for future collaborations with other researchers. Both plan to continue leveraging AI to create meaningful benefits for the American public.
“The more I study, teach and research artificial intelligence, the more I appreciate human intelligence,” Yang says. “As we continue to do this work, we must do everything we can to ensure this technology is used for social good.
