Waxaa qoray Fesax Benson • 10 Nofembar 2026
Yeruusaalem, 10 Nofembar, 2026 (TPS-IL) — Iyadoo malaayiin qof ay si isa soo taraysa ugu tiirsan yihiin sirdoonka macmalka ah (AI) si ay u fahmaan dagaallada, khilaafaadka siyaasadeed, iyo wararka degdegga ah, daraasad cusub oo Israa’iili ah ayaa su’aalo ka keenaysa aragtida cidda ay nidaamyadani muujiyaan marka ay wararka fasirayaan.
Cilmi-baarayaal ka tirsan Jaamacadda Ben-Guriyoon ee Naqab ayaa ogaaday in moodellada ugu horreeya ee AI ay si weyn ugu kala duwan yihiin sida ay u fasiraan calaamadaha shucuurta ee ay xambaarsan yihiin cinwaannada wararku. Moodellada qaarkood waxay aad ugu dhowaayeen jawaabaha akhristayaasha aadanaha ah, balse midna si joogto ah uma tarjumin aragtida dhammaan kooxaha.
Natiijooyinku waxay muujinayaan in nidaamyada AI aysan qasab ahayn inay bixiyaan fasiraad ay dhammaan wadaagaan ee ku saabsan wararka, gaar ahaan marka laga warramayo colaadaha xasaasiga ah ee siyaasadeed iyo kuwa caalamiga ah.
Daraasaddu waxay baartay sida ereyada loo adeegsado cinwaannada ay u saameyn karaan dareenka akhristayaasha. Laba cinwaan oo qeexaya isla hal dhacdo, tusaale ahaan, ayaa laga yaabaa inay xoogga saaraan dhinacyo kala duwan oo sheekada ka mid ah, iyadoo midkood uu diiradda saarayo dhibaatada dadka rayidka ah kan kalena xaaladaha milatari. Labaduba waxay noqon karaan kuwo xaqiiqo ahaan sax ah iyagoo haddana kicinaya heerar kala duwan oo naxariis ah oo loo hayo dadka ay khuseyso.
Cilmi-baarayaashu waxay doonayeen inay xaqiijiyaan bal in moodellada AI ay garteen kala duwanaanshahan iyo in fasiraaddoodu ay la socotay tan akhristayaasha aadanaha ah.
Kooxda cilmi-baarista, oo ay ku jireen Dr. Haran Shani Narkiss, Dr. Michael Fire, iyo Dr. Oren Tsur, waxay barbardhigeen toddoba moodel oo AI ah oo hormuud ah iyo jawaabaha laga helay 3,011 qof oo qaan-gaar ah oo ku nool Boqortooyada Midowday.
Ka-qaybgalayaasha iyo moodellada AI ayaa loo bandhigay cinwaanno warar ah oo ku saabsan dagaalka Ruushka iyo Yukrayn, dagaalka Gaza, iyo ololihii doorashada madaxtinnimada Maraykanka ee 2024. Waxaa la weydiiyay in cinwaan kasta uu kiciyay naxariis loo hayo dhinac cayiman. Sahanka YouGov ee matalaya heer qaran ayaa soo saaray in ka badan 200,000 oo jawaabo shaqsi ah.
“LLM-yada ayaa si isa soo taraysa u noqonaya dhexdhexaadiyeyaal u dhexeeya dadka iyo macluumaadka,” ayuu yiri Dr. Tsur, oo ah kaaliye borofisar ka tirsan Machadka Cilmi-baarista AI ee La Adeegsado ee jaamacadda.
“Saamaynta cinwaanka warku waxay dhaafsiisan tahay xaqiiqooyinka uu tebiyo. Erayada loo isticmaalay waxay gudbin karaan naxariis, si dadban ayay mas’uuliyad u saari karaan, waxayna si xeeladaysan u qaabayn karaan fahamka akhristaha,” ayuu yiri.
Cabbiridda Eexda AI
Cilmi-baarayaashu waxay ogaadeen in heerka ay moodellada AI ku raaceen jawaabaha aadanaha uu ku kala duwanaa da’da ka-qaybgalayaasha, jinsiga, waxbarashada, aqoonta hore ee arrimaha juqraafiga siyaasadeed, iyo aragtiyadooda hore ee ku aaddan colaadaha.
Si kale haddii loo dhigo, moodel cinwaannada u fasiray si la mid ah hal koox oo dad ah ma uusan ka tarjumayn aragtiyaha koox kale.
Dr. Michael Fire, oo ah bare sare oo ka tirsan jaamacadda, ayaa sheegay in baaxadda sahanku ay u suurtagelisay cilmi-baarayaasha inay baaraan xukunnada kooxaha ay moodellada AI sida ugu dhow uga tarjumeen, halkii ay kaliya ku tiirsanaan lahaayeen natiijooyinka celceliska ah.
“Iyadoo ay jiraan in ka badan 200,000 oo jawaabood oo laga helay in ka badan 3,000 oo ka-qaybgale, waxaan eegi karnaa wax ka baxsan in AI-gu uu dadka ku raacsan yahay heerka celceliska ah,” ayuu yiri Fire.
Natiijooyinku waxay wataan saameyn ballaaran iyadoo aaladaha AI ay si isa soo taraysa u noqonayaan ilo caadi ah oo laga helo soo-koobidda iyo sharraxaadda wararka. Marka nidaamka AI uu fasirayo cinwaan ama uu soo koobayo warbixin, waxaa laga yaabaa inuu seego calaamadaha dareenka ee saameeya sida akhristayaashu u fahmaan dhacdo.
Cilmi-baarayaashu waxay sheegeen in horumarinta mustaqbalka ee AI laga yaabo inay u baahato inay xisaabta ku darto kala duwanaanshaha tirakoobka dadweynaha iyo dhaqanka. Si kastaba ha ahaatee, waxay ka digeen in go’aaminta aragtida cidda ay tahay inay hagto nidaamyada AI ay dhalinayso su’aalo adag oo dhanka anshaxa ah. Haddii aanay jirin hufnaan, dadaallada lagu waafajinayo moodellada dhagaystayaal gaar ah waxay horseedi karaan eex cusub.
Waxay sidoo kale ka digeen in nidaamyada AI ee si joogto ah ula raaca isticmaalayaasha isla markaana xoojiya fikradahooda jira ay gacan ka geysan karaan abuurista goobo ay fikradaha isku midka ahi isku sii celiyaan, kordhinta kala qaybsanaanta, iyo xaalufinta kalsoonida dadweynuhu ku qabaan ilaha macluumaadka.