Energy Aims To Make AI Human-Driven, Accessible To Underserved Communities
Westend61/Getty Images
Having real people support the artificial intelligence system both builds community trust and ensures that those systems are designed for underserved populations.
Officials within the Department of Energy are looking to apply practical uses of artificial intelligence technology to helping underserved communities.
Speaking during a FedScoop discussion panel, Pamela Isom, the director of the Artificial Intelligence and Technology Office at Energy, explained the importance of using AI technology to strategically help, as it becomes more and more ubiquitous in daily life.
Some current use cases for AI tech within Energy are automated loan and application processing. Ipsom elaborated that her office’s mission to ensure equitable access to AI technology came from a community discussion where gaps in adequate technological infrastructure were highlighted.
“Today, we're looking at AI for instance, to not only help with procurement cycles, but with processing and evaluating [requests for information] for instance,” Isom said. She noted that natural language processing can both save time and catch errors on a digital imaging level that may not have otherwise been caught if reviewed manually.
“AI is everywhere,” Isom said, explaining that common use cases should think about underserved communities and their needs. “Our focus as a department is on impact.”
In order to achieve this, Isom highlighted the need for adequate research and development efforts to hone in on what challenges the technology should work on addressing.
Isom also said that improving data availability and labeling is also crucial to generating positive outcomes with AI, but noted the human element behind AI applications is “so important.”
“I think that right now we need humans in the loop because of AI and the state that it's in. There's opportunities to strengthen that confidence in citizens. And so the human is in the loop to verify and validate,” she said.
Keeping a people-centric design in the early stages of AI technology development will help build confidence in AI products and systems among the general public. Isom emphasized the importance of detailed understanding of data and datasets among employees developing AI technologies as another key factor in deploying unbiased machine learning systems running on diversified information and data.
“The greater the data represents, the greater the likelihood of generating outcomes that will have the impact that we so desire,” she said.
Over time, the relationship between AI systems and the humans behind them is likely to change, according to Isom, but that won’t happen until the technology matures significantly. Due to this prerequisite, she advocated the development of a robust workforce ready to implement AI technology.
Isom’s comments come as more federal agencies seek to implement AI in standard business processes, but are focusing on the workforce and human element that programs the technology itself.
Last June, the Biden administration launched the National Artificial Intelligence Research Resource Task Force, which supports both researching and implementing AI in federal offices, as well as monitoring the societal and security impact.