{"id":166548,"date":"2026-03-27T17:12:55","date_gmt":"2026-03-27T15:12:55","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/greg-kroa-hartman-rasskazal-chto-llm-stali-luchshe-iskat-oshibki"},"modified":"2026-03-27T17:12:55","modified_gmt":"2026-03-27T15:12:55","slug":"greg-kroa-hartman-rasskazal-chto-llm-stali-luchshe-iskat-oshibki","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/news\/greg-kroa-hartman-rasskazal-chto-llm-stali-luchshe-iskat-oshibki","title":{"rendered":"Greg Kroah-Hartman theksoi se LLM-t\u00eb kan\u00eb filluar t\u00eb gjejn\u00eb m\u00eb mir\u00eb gabimet","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
Greg Kroah-Hartman, nj\u00eb zhvillues holandez i Linux-it, i tregoi The Register p\u00ebr ndryshimet n\u00eb raportet e dozave t\u00eb m\u00ebdha t\u00eb vulnerabiliteteve, t\u00eb gjeneruara nga 'intelgjenca artificiale'.<\/p>\n
Nj\u00eb muaj m\u00eb par\u00eb, cil\u00ebsia e raporteve t\u00eb dob\u00ebsive nga AI u p\u00ebrmir\u00ebsua ndjesh\u00ebm. M\u00eb par\u00eb, shumica e tyre ishin rrenja, por sipas Kroah-Hartmanit, kjo nuk ishte nj\u00eb problem i madh p\u00ebr b\u00ebrtham\u00ebn (ndryshe nga cURL) \u2014 ka shum\u00eb zhvillues dhe ata po e menaxhojn\u00eb, nd\u00ebrsa rrenjat ishin m\u00eb shum\u00eb qesharake se sa shqet\u00ebsuese. Por n\u00eb muajin e fundit nj\u00eb pjes\u00eb e konsiderueshme e raportuar ka qen\u00eb v\u00ebrtet e dobishme. Dhe kjo v\u00ebrehet edhe n\u00eb projekte t\u00eb tjera t\u00eb hapura. Pse \u00ebsht\u00eb k\u00ebshtu, nuk dihet. Qoft\u00eb mjetet jan\u00eb p\u00ebrmir\u00ebsuar, ose njer\u00ebzit kan\u00eb filluar t'i p\u00ebrdorin ato si\u00e7 duhet.<\/p>\n
Zhvilluesit q\u00eb d\u00ebrgojn\u00eb raporte dhe patch-e fillojn\u00eb ta etiketojn\u00eb ndershm\u00ebrisht at\u00eb q\u00eb \u00ebsht\u00eb b\u00ebr\u00eb nga AI si t\u00eb till\u00eb. AI tani \u00ebsht\u00eb n\u00eb gjendje t\u00eb gjeneroj\u00eb dhjet\u00ebra patch-e, n\u00ebse kushtet p\u00ebr shfaqjen e gabimeve jan\u00eb t\u00eb thjeshta dhe leht\u00ebsisht t\u00eb identifikueshme.<\/p>\n
P\u00ebrsa i p\u00ebrket Sashiko t\u00eb d\u00ebrguar nga Google, Kroah-Hartman tha se ata kan\u00eb nevoj\u00eb p\u00ebr nj\u00eb mjet p\u00ebr vler\u00ebsimin e shpejt\u00eb t\u00eb patch-eve t\u00eb d\u00ebrguara. Shumica e patch-eve tani kalojn\u00eb p\u00ebrmes tij. Dhe ai \u00ebsht\u00eb n\u00eb dispozicion p\u00ebr t\u00eb gjith\u00eb.<\/p>\n
LLM p\u00ebr revizionet u p\u00ebrdor\u00ebn p\u00ebr her\u00eb t\u00eb par\u00eb nga zhvilluesit e n\u00ebnstruktur\u00ebs rrjet dhe BPF. M\u00eb pas DRM. Por kjo k\u00ebrkonte shum\u00eb burime. Me mjetin e Google, kjo \u00ebsht\u00eb b\u00ebr\u00eb e disponueshme p\u00ebr t\u00eb gjith\u00eb, dhe gjith\u00e7ka do t\u00eb shkoj\u00eb n\u00eb nj\u00eb nd\u00ebrfaqe t\u00eb p\u00ebrbashk\u00ebt.<\/p>\n
K\u00ebto verifikime jan\u00eb 'shtes\u00eb' p\u00ebr ato ekzistuese, e jo 'p\u00ebrfundimtare'. Ato lejojn\u00eb p\u00ebrgjigje m\u00eb t\u00eb shpejta p\u00ebr zhvilluesit q\u00eb d\u00ebrgojn\u00eb patch-e. Kroah-Hartman pret q\u00eb numri i raporteve p\u00ebr gabime dhe patch-eve t\u00eb krijuar nga AI-t\u00eb do t\u00eb rritet, dhe po k\u00ebrkon m\u00ebnyra p\u00ebr t\u00eb p\u00ebrshpejtuar p\u00ebrpunimin e tyre. N\u00ebse di\u00e7ka nuk kalon verifikimin automatizues, kujdestari mund t\u00eb mos p\u00ebrqendrohet p\u00ebr t\u00eb.<\/p>\n
Kroah-Hartman beson se AI ka leht\u00ebsuar gjetjen e dob\u00ebsive dhe ka rritur ngarkes\u00ebn p\u00ebr mir\u00ebmbajt\u00ebsit e b\u00ebrtham\u00ebs. Prandaj do t\u00eb ishte e drejt\u00eb q\u00eb t\u00eb leht\u00ebsohet puna e mir\u00ebmbajt\u00ebsve me ndihm\u00ebn e AI.<\/p>\n