SEATTLE — Every day, we hear more ways artificial intelligence is being integrated into our lives.
We have seen AI for years, like when Netflix has an automated list of suggested movies or shows. But what is emerging now is generative AI.
RELATED: OpenAI opens Bellevue office with ribbon-cutting as region reels from recent tech layoffs
“It creates more opportunities for people to be able to engage in various ways,” said Dr. Jazette Johnson, a postdoctoral fellow at UW.
And a lab at the University of Washington is looking at new ways to use or improve the technology.
“It is important that not pause in development and engagement but also bring caution when we are presenting various things,” Johnson said.
Johnson is homing in on systems to help adults with dementia or other cognitive disabilities.
“Now [patients] will go to ChatGPT and say, ‘Hey, tell me what that means,’ but you don’t have boundaries with ChatGPT right now,” Johnson said. “So, I created a system where you can actually have those boundaries. If it’s something that’s about diagnoses and complicated information, go talk to your doctor, but if I’m just trying to understand a lab result, here’s the information.”
Johnson said the idea is both to have the patient information in the backend, so it is all accurate, and then about how “we can use AI to simplify language” that often appears on online patient portals and charts.
And the program’s prototype has already been tested by 8 adults with impairment and one caregiver for someone with dementia.
“When they actually engaged with the specific tool, they were like ‘wow, I did not know this could happen,'” Johnson said.
Johnson said as the program is further developed, of course, implementation of things like HIPAA will need to be at the forefront of mind. But she said overall, one day the program could grow and eventually be used by anyone in their patient portal.
If you are interested in learning more about AI literacy, you can sign up for a workshop hosted by Johnson here.
RELATED: Parents and kids weigh in on AI’s impact, from schoolwork to future job opportunities
Kate Glazko is a third-year PHD student at UW’s Allen Center of Computer Sciences; she identifies as someone with a disability.
“People with disabilities or health conditions can be playing an active role in their own accessibility now,” Glazko said.
Glazko’s work studies several facets of generative AI. One of her first research projects in the area looked at how AI resume sorters put people with disability-related rewards or scholarships at a disadvantage, using what she calls biased language to describe that experience.
“For example, it would describe the resume, with additional positive autism items, including a scholarship and a reward, as less leader-like, even though those resumes had identical leadership experience,” Glazko explained.
So, Glazko said her group created a version of ChatGPT that was trained not to be biased against disability-related accolades.
“Which gave us hope,” she said.
But, despite attention, Glazko said the new techniques simply were not being embraced, so they pivoted in the development of that AI tech.
“We started to think about how we can address these issues from the job seeker’s side,” she said.
Now, what’s being studied are systems to help people tailor their resumes to ensure they are not sorted out for things like disability-related awards, while keeping their identity. The technology is not widely available yet, Glazko said, but is being studied.
RELATED: AI use on the job rising, but fast adoption faces speed bumps
Glazko is also working on technologies to help people create personalized reminder bracelets, for example.
“That is the power of generative AI, when it comes to accessibility, it opens doors for people to build their own solutions and personalized solutions, too,” she said.
