Improving Employee Satisfaction by Asking the Right AI Questions
Amazon also has AI using incident response tooling previously operated by people. The company’s inability to hire enough talent with the proper skill sets, an industrywide problem, drove the decision.
“Systems tend to do what you tell them to do; they do it correctly the first time, every time, usually,” said Stephen Schmidt, CSO at Amazon, on the same summit panel. “Turns out, our staff is actually continually happier with their jobs at the end of the day if we remove a lot of that front work from them.”
Happy employees are employees you can retain, a win for agencies also struggling with hiring enough tech talent.
Schmidt recommends that organizations first ensure proper access controls are in place for any data AI might touch, and they should ask themselves a crucial question: Should AI be doing this? If the answer is yes, getting the technology into end users’ hands quickly and safely often leads to them discovering new ways to use the models faster.
RELATED: Machine learning models are expediting federal tech efforts.
There are other good questions to ask at this point: Why can’t AI do this? When will it be able to? Under what conditions?
“If they’re not intentionally making that decision, they’re missing an opportunity,” Schmidt said. “And all of us will pay by being late later on.”
The CIA Eyes Agentic AI
The CIA is looking for hires who live at the intersection of technology and human behavior, analysts who think critically with incomplete information while demonstrating expertise and curiosity about AI, languages and regions, Raman said.
Agentic AI — a new form capable of taking actions on behalf of people, such as making a reservation — opens the door to performing complex, multistep workflows or tool calling across databases. Raman is most interested in those applications, as well as business use cases such as the CIA’s help desk and form filling.
UP NEXT: Agencies must be proactive in securing data from artificial intelligence threats.
Still, she has reservations about agentic AI.
“You have to be concerned about drift happening, Raman said. “What’s going on inside of that black box, and do you have a level of explainability around that for your users?”
