Page 6 - FCW, September/October 2019

Trending
AI to streamline
past-performance
data
The Department of Homeland Security released a solicitation in August call- ing for an artificial intelligence-based tool that can help contracting officials rapidly pull relevant information on past-performance evaluations for a given contractor or vendor from the Contractor Performance Assessment Reporting System.
The tool would help streamline the federal contracting process and admin- istrative workloads.
DHS seeks a proof of concept that uses advanced automation, continual learning, prediction and decision aids to solve complex, dynamic problems. The AI must be “human readable,” meaning people can understand how the AI delivers results, and verifiable, meaning it produces objective results “that can be trusted, are free from biases, and that produce intended outcomes in a repeatable and objec- tive manner.”
Besides using AI to streamline CPARS evaluations, DHS also wants data-driven recommendations about improving the quality of the past- performance data that contracting offi- cers enter into the system.
DHS will award up to $50,000 to ven- dors that can provide a prototype that successfully delivers relevant informa- tion based on test data. Follow-on con- tracts may be awarded to prototypes that use AI to “predict whether the offeror being evaluated will successful- ly perform the future contract based on its past-performance record in CPARS, without further competition.”
The ultimate goal of the AI initiative is a “self-sustaining multi-contractor commercial marketplace, perhaps simi- lar to the credit report marketplace,” the solicitation states.
— Anoushka Deshmukh
3 years is the sole-source extension Immigration and Customs Enforcement awarded to Palantir for case management and analytics
Army enlists AI in search of better batteries
Warfighters increasingly rely on electronic gear in the field, so researchers are looking for ways to make the batteries that power such equipment safer, lighter and less expensive.
The Army Research Laboratory’s Army Research
Office is funding a team at
Cornell University that is
using artificial intelligence
to identify alloys that would contribute to more efficient
energy storage. Multiple bots
took on different aspects
of the alloy identification
problem, “from predicting the
phase structures of various combinations to making sure
those predictions obey the
rules of thermodynamics,” according to an ARO news release.
The system helped researchers identify a unique catalyst that could be incorporated into more efficient fuel cells, and it was partly inspired by the IBM Watson supercomputer that called on a community of AI agents to answer questions on the “Jeopardy!” quiz show.
“Power is absolutely important in the Army,” said Purush Iyer, chief of ARO’s Network Sciences Division. “Soldiers walk from place to place carrying their gear. They have tons of equipment with
them \[and\] tend to carry very expensive batteries. It can be a limiting factor. Portable, safe fuel cells are definitely incredibly important.” He added that such fuel cells could also be beneficial
in electric cars and other nondefense applications.
Iyer said officials hope that extended- life battery technology will be deployable in the next five to 10 years, though it is hard to pinpoint a timeline for field delivery. But he stressed that the Defense Department and society in general must manage their expectations for AI.
“Machine learning is not a silver bullet,” Iyer said. “You have to apply it in a very, very refined way to get the kind of benefits that are necessary.”
— Lauren C. Williams
Dorris Consulting International
@DorrisConsultng
CX is more important now than ever in the fed sector. As quoted in the article below, 80% of consumers say the experience an organization provides is just as important as its products and services. @FCWnow @salesforce
10:26 AM · Aug 19, 2019
Reply Revtweet
Favorite
Join the conversation
FCW usesTwitter to break news, field questions and ask our own.
Learn more atTwitter.com/FCWnow.
6
September/October 2019 FCW.COM