USA

The agency advised to approach AI from an open and collaborative mindset

In Zach Goldfine’s view, the fact that it took veterans about 100 days to get the news that help was coming to process a benefit claim was conscientious, not to mention the actual time to get help. It wasn’t.

But that was the reality before the Department of Veterans Affairs Release Last year’s Content Classification Prediction Service Application Programming Interface. More than 1.5 million claims for disability compensation and benefits are filed each year, 65% -80% of which are mailed or faxed. And 98.2% of attempts to automate the language of those claims were unsuccessful.

“Veterans can write anything, but they think of the injuries they have suffered, so they might say,’My ears hurt and there is a constant ringing sound’ … but VA says,” I My ears hurt and I’m constantly ringing “doesn’t benefit-it either benefits or gives a monthly payment for hearing loss.” Goldfine, Deputy Chief Technology Officer of Benefits at VA, said. “The problem was that veterans faced an additional five days delay in making decisions about their interests, because there was a back-up of claims in that part of the process that needed to make that translation. Because. “

After deploying the API, Goldfine said that almost overnight, the number of successfully automated claims tripled, saving millions of dollars from reducing the time required to translate claims.This year, VA was deployed rapidly, reducing latency again. Chatbot Answer questions about COVID-19, such as which facilities are open.

This was an example of a successful artificial intelligence institution highlighted during the Impact Summit Series: Artificial Intelligence, announced Thursday by the Department of Technology Transformation Services of the General Procurement Department. According to Goldfine, the API case at VA shows how AI makes it easier for employees to do their jobs, rather than obsolete them. This is a common fear that organizations and managers are wary of implementing AI in their offices. Expect to talk to staff and hear various concerns between managers and lower-level workers performing these corrupt tasks.

“We all have some things we don’t like, like the one million emails we need to reply to, entering certain data elements when we need to take a vacation. Work is always monotonous. There’s a part. What is our job and I think most people want to be able to automate them whenever possible, “he said. “In scenarios where many of them were veterans, they manually re-entered the data. They talked to veterans on the phone, understood what was happening, and got the information they needed. And so on, we want to spend our time doing what we want to do. “

Goldfine also emphasized building an interdisciplinary team to implement AI, gain support, and consider its impact on humans. System designers, product managers, user researchers, data scientists, software engineers, and even policy professionals are all important.

Alka Patel, Head of Pentagon AI Ethics Policy The Joint Artificial Intelligence Center brings engineering and legal background to her role. She said good engineering principles of design, development, deployment, and use are combined with risk management and government and corporate compliance considerations.

When DoD adopts AI ethical principles During February – After two years of work leading up to that, Patel’s responsibility is “Take those higher level words and definitions and actually make them tactical. “

When it comes to ethical AI, her advice to the agency was to start now and use existing AI strategies as a framework. Also, it may not be possible to predict all possible scenarios and ethical issues, but as a general rule, as technology evolves, the testing and evaluation process becomes more variable. , Some things probably remain solid.

It’s better to think of ethics as a realization factor for AI rather than an obstacle. In addition, the award that contractors need to comply with DoD AI ethical principles can be helpful from a signaling perspective, but Patel was skeptical that it would serve its desired purpose.

“I am very sensitive to’instructing’that these requirements need to be implemented from an agency perspective. I think it’s a conversation that needs to be done with our contractors, or at least have some insight, “she said. “We don’t have to be very normative, we need flexibility, but we still need to have content fidelity and the standards we expect from the contractors themselves.”



The agency advised to approach AI from an open and collaborative mindset

Source link The agency advised to approach AI from an open and collaborative mindset

Back to top button