Warbixin GAO oo Oktoobar la daabacay ayaa sheegtay in qaar ka mid ah saraakiisha ay aqoonsadeen khataro iyo in aan la hubin la-socodka mustaqbalka.
Warbixin ay soo saartay xafiiska GAO ee Maraykanka 14-kii Oktoobar ayaa sheegtay in Pentagon uu dhisayo hub iyo mashaariic AI ah oo aan had iyo jeer si dhow u raacin tilmaamaha hay'adda ee AI-da mas'uulka ah, sida ay saraakiil sheegeen.
Khataraha la aqoonsaday iyo baaxadda mashaariicda
GAO waxay waraysatay 10 sarkaal oo ka qayb qaatay dhisidda 10 hub iyo qalab oo Pentagon u isticmaalo AI si ay si otomaatig ah u aqoonsadaan bartilmaameedyada dagaalka, u farsameeyaan sahaminta iyo u wanaajiyaan go'aannada taktigga. Mashaariicdani waxay ka mid yihiin in ka badan 300 oo AI lagu dhisayo Pentagon. Lix ka mid ah 10-ka sarkaal ayaa sheegay in kooxahooda ay aqoonsadeen khataro ku xiran isticmaalka mashaariicda deegaan hawleed.
Tilmaamaha 2021 iyo ku-dhaqanka
Sannadkii 2021, Pentagon waxay soo dirtay memo lagu qeexayo lix mabda' oo aasaasi ah si hubka AI lagu dhiso si mas'uul ah. Mabda'ahan waxay xoogga saaraan kormeerka guud ee Pentagon, xaqiijinta iyo la-socodka waqtiga dhabta ah, iyo iskaashiga xulafada. Alissa Czyz oo hagaysay dib-u-eegista GAO waxay sheegtay in mabda'ahan ay ka soo jeedeen mabaadi' hore, inkastoo aanay xeerar ahayn. Afar sarkaal ayaa GAO u sheegay in shuruudaha mashaariicdooda aanay waafaqsanayn mabaadi'da AI-da mas'uulka ah.
La-socodka mustaqbalka iyo shaqaalaha
Saddex ka mid ah 10-ka sarkaal ayaa GAO u sheegay in aanay garanayn in mashaariicda si joogto ah lala socon doono marka la dhoofiyo. Qaybta AI-da mas'uulka ah ayaa yaraaday shaqaale: 2023 waxaa loo oggolaaday inay qortaan 10 qof, laakiin bishii Febraayo saddex keliya ayaa buuxsamay. GAO waxay soo jeedisay laba talo, oo Pentagon ay ogolaatay, laakiin ilaa 22 Oktoobar aan la fulin.
A congressional watchdog surveyed 10 officials building AI-enabled weapons and tools; six reported identified risks in operational use.
A U.S. Government Accountability Office report published Oct. 14 found that Pentagon AI projects do not always closely follow the agency's own responsible AI guidelines, according to officials surveyed. Six of 10 officials said their teams had "identified risk(s)" with using the projects "in an operational environment."
What the GAO Survey Found
The GAO surveyed 10 officials involved with building 10 weapons and tools that use AI to identify targets, process surveillance and improve tactical decision making, among other tasks. The projects are among more than 300 being built with AI at the Pentagon. Four officials said requirements for their projects were not "aligned with Responsible AI tenets," and three of the 10 said they "don't know" whether projects will be continuously monitored after deployment.
Guidelines Are a Memo, Not a Department Instruction
In 2021, the Pentagon issued a memo outlining six tenets stressing Pentagon-wide oversight, verification, "real-time monitoring" and cooperation with allies "to advance global norms grounded in shared values." Alissa Czyz, who directed the GAO review, said the tenets were understood as important but were not rules. "It's a memo, right, at the end of the day, so it's not a department instruction," Czyz said.
Understaffing and Unimplemented Recommendations
The Department of the Responsible AI Division, which helps manage education on the guidelines, was authorized to hire 10 people in 2023, but only three positions were filled as of February, and the Pentagon "is unlikely to fill vacant positions or increase staffing," the report states. Czyz said: "That's a very small number of folks tasked with a large responsibility of responsible AI throughout the department." The GAO made two recommendations; the Pentagon agreed but had not implemented them as of Oct. 22.
Ilaha iyo xuquuqda sawirka
Sawir: NPR News Xigasho
