{"id":98121,"date":"2026-05-13T09:03:59","date_gmt":"2026-05-13T07:03:59","guid":{"rendered":"https:\/\/digitalskills.mdia.gov.mt\/article\/nghixu-bmudelli-lingwistici-kbar\/"},"modified":"2026-05-13T09:03:59","modified_gmt":"2026-05-13T07:03:59","slug":"nghixu-bmudelli-lingwistici-kbar","status":"publish","type":"article","link":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/","title":{"rendered":"Ng\u0127ixu b’Mudelli Lingwisti\u010bi Kbar"},"content":{"rendered":"

Astratt<\/h2>\n

Mudelli lingwisti\u010bi kbar m’g\u0127adhomx limitati g\u0127al-laboratorji tar-ri\u010berka. Huma integrati f’applikazzjonijiet tal-konsumatur u pjattaformi tax-xog\u0127ol, u issa jinfluwenzaw kif in-nies jiktbu, ifittxu, jikkodifikaw, u jie\u0127du de\u010bi\u017cjonijiet. Dan l-artiklu jispjega, f’termini sempli\u010bi i\u017cda rigoru\u017ci, kif ja\u0127dmu s-sistemi kontemporanji tal-IA \u0121enerattiva, g\u0127aliex l-arkitettura tat-trasformaturi kienet kru\u010bjali, u x’inbidel meta l-i\u017cvilupp tal-mudell inbidel minn ri\u010berka miftu\u0127a g\u0127al kompetizzjoni kummer\u010bjali. Jiddeskrivi wkoll implikazzjonijiet so\u010bjetali ewlenin, inklu\u017ci l-edukazzjoni, is-swieq tax-xog\u0127ol, il-kreattivit\u00e0, u l-governanza. <\/p>\n

Il-mistoqsija essenzjali dan l-a\u0127\u0127ar mhijiex jekk il-magni jistg\u0127ux ja\u0127sbu. Hija jekk in-nies jistg\u0127ux jaffordjaw li ju\u017cawhom ming\u0127ajr ma jifhmu x’inhuma. <\/em><\/p>\n

Introduzzjoni<\/h2>\n

L-intelli\u0121enza artifi\u010bjali mxiet minn su\u0121\u0121ett spe\u010bjalizzat g\u0127all-\u0127ajja ta’ kuljum. G\u0127odod b\u0127al ChatGPT ta’ OpenAI, Gemini ta’ Google, Claude ta’ Anthropic, u Microsoft Copilot jintu\u017caw \u0127afna biex jabbozzaw test, ji\u0121bru fil-qosor l-informazzjoni, jiktbu softwer, u jappo\u0121\u0121jaw komunikazzjoni ta’ rutina. Il-konvenjenza tag\u0127hom tista’ ta\u0127bi realt\u00e0 importanti: \u0127afna utenti jintera\u0121ixxu ma’ dawn is-sistemi b\u0127allikieku kienu interlokuturi infurmati, filwaqt li jkollhom biss fehim limitat tal-g\u0127a\u017cliet tekni\u010bi u organizzattivi li jsawru l-outputs tag\u0127hom. <\/p>\n

Mir-Ri\u010berka g\u0127all-Adozzjoni tal-Massa<\/h2>\n

\"Ng\u0127ixu<\/a><\/p>\n

L-ewwel de\u010bennji tal-IA kienu definiti minn avvanzi intermittenti u perjodi twal ta\u2019 progress inkrementali. Il-mew\u0121a attwali hija differenti g\u0127aliex tg\u0127aqqad tliet kundizzjonijiet: kwantitajiet vasti ta\u2019 test di\u0121itali, cloud computing skalabbli, u tekniki ta\u2019 tag\u0127lim awtomatiku li jitjiebu b\u2019mod prevedibbli bl-iskala. Ir-ri\u017cultat huwa klassi ta\u2019 sistemi li jistg\u0127u ji\u0121\u0121eneraw lingwa\u0121\u0121 fluwenti u, f\u2019\u0127afna kuntesti, spjegazzjonijiet, sommarji u kodi\u010bi utli. <\/p>\n

Fl-istess \u0127in, i\u010b-\u010bentru tal-gravit\u00e0 mxiet minn pubblikazzjoni akkademika miftu\u0127a lejn \u017cvilupp kummer\u010bjali. \u0126afna organizzazzjonijiet ewlenin g\u0127adhom jippubblikaw ri\u010berka, i\u017cda qed jillimitaw dejjem aktar l-i\u017cvelar ta\u2019 dejta dwar it-ta\u0127ri\u0121, dettalji tad-disinn tas-sistema, u metodi ta\u2019 ottimizzazzjoni. Dan huwa parzjalment rispons g\u0127all-pressjoni kompetittiva u t-t\u0127assib dwar il-proprjet\u00e0 intellettwali, u parzjalment riflessjoni ta\u2019 kunsiderazzjonijiet ta\u2019 sikurezza u sigurt\u00e0. G\u0127all-qarrejja, il-konsegwenza prattika hija li l-fehim ta\u2019 dawn is-sistemi issa je\u0127tie\u0121 li jing\u0127aqdu r-ri\u010berka pubblika ma\u2019 interpretazzjoni bir-reqqa ta\u2019 dak li l-kumpaniji jag\u0127\u017clu li ji\u017cvelaw. <\/p>\n

Kif Mudelli Lingwisti\u010bi Kbar Ji\u0121\u0121eneraw Test<\/h2>\n

Il-bi\u010b\u010ba l-kbira tas-sistemi kontemporanji bba\u017cati fuq i\u010b-chat huma mibnija fuq mudelli lingwisti\u010bi kbar (LLMs). LLM huwa m\u0127arre\u0121 fuq corpora ta’ test massivi u jitg\u0127allem ibassar it-token li jmiss f’sekwenza. Token tipikament huwa bi\u010b\u010ba test qasira, \u0127afna drabi kelma, parti minn kelma, jew simbolu. Matul it-ta\u0127ri\u0121, il-mudell ripetutament jara sekwenzi ta’ tokens u ji\u0121i penalizzat meta jassenja probabbilt\u00e0 baxxa lit-token li fil-fatt isegwi. Matul \u0127afna iterazzjonijiet, dan l-objettiv awtosorveljat jinkora\u0121\u0121ixxi lill-mudell biex jinternalizza regolaritajiet statisti\u010bi li jirriflettu l-grammatika, l-istil, u \u0127afna asso\u010bjazzjonijiet fattwali pre\u017centi fid-dejta. <\/p>\n

Kliem, Im\u0127u\u0127, u Magni: Il-Mira\u0121\u0121 tal-‘Fehim’ fid-Dibattiti tal-AI<\/h2>\n

\"Ng\u0127ixu<\/a><\/p>\n

Id-dibattiti dwar dawn is-sistemi spiss jiddependu fuq kliem b\u0127al \u0127sieb, fehim, ra\u0121unament, u kuxjenza. Dawn it-termini mhumiex definiti sew g\u0127ax m’g\u0127andhomx tifsira wa\u0127da miftiehma fil-filosofija, il-psikolo\u0121ija, in-newroxjenza, u x-xjenza tal-kompjuters, u g\u0127ax jistg\u0127u jirreferu g\u0127al tipi differenti ta’ evidenza. Fid-diskors ta’ kuljum spiss jiddeskrivu m\u0121iba esterna, b\u0127as-soluzzjoni ta’ problemi jew l-ispjegazzjoni ta’ de\u010bi\u017cjonijiet, filwaqt li fl-u\u017cu xjentifiku u filosofiku jistg\u0127u jirreferu g\u0127al mekkani\u017cmi interni jew esperjenza su\u0121\u0121ettiva. Peress li m’hemm l-ebda test ta’ kunsens li jissepara b’mod \u010bar dawn it-tifsiriet, id-diskussjonijiet jistg\u0127u ji\u017c\u017cer\u017cqu bejniethom u jo\u0127olqu l-impressjoni li lingwa fluwenti timplika stati mentali simili g\u0127al dawk tal-bniedem. <\/p>\n

Dan l-objettiv ta’ ta\u0127ri\u0121 jista’ jipprodu\u010bi outputs li j\u0127ossuhom konversazzjonali u, xi drabi, insightful. Madankollu, it-tbassir ta’ next-token ma jimplikax kuxjenza, intenzjonijiet, jew esperjenza su\u0121\u0121ettiva. Il-mudell ji\u0121\u0121enera test billi jittrasforma kuntest ta’ input f’distribuzzjoni ta’ probabbilt\u00e0 fuq kontinwazzjonijiet possibbli. Meta rispons jidher li jinkludi ra\u0121unament, dan jinftiehem l-a\u0127jar b\u0127ala mudell ta’ lingwa li l-mudell tg\u0127allem jirriprodu\u010bi g\u0127aliex jikkorrelata ma’ kontinwazzjonijiet ta’ su\u010b\u010bess fid-dejta tat-ta\u0127ri\u0121 tieg\u0127u u rfinar sussegwenti. <\/p>\n

L-Avvanz tat-Trasformatur<\/h2>\n

L-LLMs moderni huma tipikament ibba\u017cati fuq netwerks newrali tat-trasformaturi, introdotti fl-2017 fid-dokument Attention Is All You Need<\/em> . It-trasformatur issostitwixxa l-ippro\u010bessar strettament sekwenzjali b’awtoattenzjoni, mekkani\u017cmu li jippermetti lill-mudell ji\u017cen ir-relazzjonijiet fost it-tokens f’passa\u0121\u0121 b’mod parallel. Dan id-disinn g\u0127amilha aktar fa\u010bli biex ji\u0121u mmudellati dipendenzi fuq medda twila ta’ \u017cmien fit-test u biex ji\u0121u m\u0127arr\u0121a mudelli kbar b’mod effi\u010bjenti fuq a\u010b\u010beleraturi moderni. <\/p>\n

Transformer fih \u0127afna parametri numeri\u010bi a\u0121\u0121ustabbli, \u0127afna drabi msej\u0127a pi\u017cijiet. Matul it-ta\u0127ri\u0121, metodi ta’ ottimizzazzjoni b\u0127al dixxendenza tal-gradjent u backpropagation ja\u0121\u0121ustaw dawn il-pi\u017cijiet biex inaqqsu l-i\u017cbalji ta’ tbassir f’biljuni ta’ e\u017cempji. Hekk kif dan il-pro\u010bess jikkonver\u0121i, il-mudell ji\u017cviluppa rappre\u017centazzjonijiet interni li jappo\u0121\u0121jaw firxa ta’ m\u0121iba downstream, inklu\u017c is-sommarju, it-traduzzjoni, u l-\u0121enerazzjoni tal-kodi\u010bi. Ir-ri\u010berkaturi g\u0127adhom jiddibattu l-a\u0127jar mod kif jinterpretaw dawn ir-rappre\u017centazzjonijiet u g\u0127aliex \u010berti kapa\u010bitajiet jidhru f’daqqa hekk kif il-mudelli ji\u017cdiedu. <\/p>\n

Ta\u0127ri\u0121 fuq Skala u l-Katina tal-Provvista tal-Kompjuters<\/h2>\n

It-ta\u0127ri\u0121 ta’ mudelli tal-fruntiera je\u0127tie\u0121 ri\u017corsi komputazzjonali sostanzjali. \u0120er\u017cijiet kbar \u0121eneralment jiddependu fuq gruppi ta’ a\u010b\u010beleraturi spe\u010bjalizzati, \u0127afna drabi GPUs, konnessi permezz ta’ netwerks b’bandwidth g\u0127oli. Bejjieg\u0127a b\u0127al NVIDIA jimmanifatturaw \u0127afna mill-a\u010b\u010beleraturi u\u017cati f’dawn il-gruppi, u fornituri kbar tal-cloud jintegrawhom f’\u010bentri tad-dejta ddisinjati g\u0127al komputazzjoni parallela intensiva. F’din l-iskala, id-disponibbilt\u00e0 tal-hardware, il-konsum tal-ener\u0121ija, it-tkessi\u0127, u n-netwerking isiru restrizzjonijiet li jinfluwenzaw liema organizzazzjonijiet jistg\u0127u j\u0127arr\u0121u u joperaw l-akbar mudelli. <\/p>\n

Il-kapa\u010bit\u00e0 tal-mudell tjiebet ukoll permezz ta’ metodi lil hinn mill-iskala mhux ippro\u010bessata. L-irfinar jadatta mudell \u0121enerali g\u0127al kompiti spe\u010bjalizzati jew kuntesti organizzazzjonali. Il-\u0121enerazzjoni awmentata bl-irkupru tg\u0127aqqad l-outputs tal-mudell ma’ tfittxijiet fuq kollezzjonijiet esterni, li jistg\u0127u jtejbu l-fattwalit\u00e0 meta s-sorsi rkuprati jkunu affidabbli. \u0126afna sistemi huma wkoll multimodali, ji\u0121ifieri jistg\u0127u jippro\u010bessaw kombinazzjonijiet ta’ test, imma\u0121ni, awdjo, jew kodi\u010bi, g\u0127alkemm il-komponent ewlieni tal-lingwa xorta tipikament jistrie\u0127 fuq immudellar ibba\u017cat fuq transformer. <\/p>\n

Ftu\u0127, Proprjet\u00e0 Intellettwali, u Trasparenza<\/h2>\n

\"Ng\u0127ixu<\/a><\/p>\n

Storikament, \u0127afna progress fl-IA \u0121ie dokumentat permezz ta\u2019 dokumenti riveduti mill-pari u implimentazzjonijiet open-source. Dik il-prattika tkompli, spe\u010bjalment fl-universitajiet u f\u2019partijiet mir-ri\u010berka tal-industrija. Madankollu diversi sistemi kummer\u010bjali ta\u2019 prestazzjoni g\u0127olja ji\u0121u rilaxxati b\u2019dettall tekniku limitat. Id-dejta tat-ta\u0127ri\u0121 \u0127afna drabi ma ti\u0121ix \u017cvelata, il-pi\u017cijiet tal-mudell jistg\u0127u jin\u017cammu, u l-mekkani\u017cmi relatati mas-sikurezza jistg\u0127u ji\u0121u deskritti biss f\u2019livell g\u0127oli. Dan inaqqas l-iskrutinju indipendenti u jag\u0127milha aktar diffi\u010bli g\u0127al dawk minn barra biex jivvalutaw il-pre\u0121udizzju, il-proprjetajiet tas-sikurezza, u l-provenjenza tal-materjal tat-ta\u0127ri\u0121. <\/p>\n

L-i\u017cvelar ristrett mhux biss dwar il-protezzjoni tal-vanta\u0121\u0121 kummer\u010bjali. L-i\u017cviluppaturi jiffa\u010b\u010bjaw ukoll t\u0127assib reali dwar l-u\u017cu \u0127a\u017cin, inklu\u017c phishing awtomatizzat, \u0121enerazzjoni ta’ malware, u informazzjoni \u017cbaljata fuq skala kbira. L-isfida g\u0127all-politika hija li t\u0127e\u0121\u0121e\u0121 it-trasparenza li tippermetti r-responsabbilt\u00e0 u l-evalwazzjoni xjentifika, filwaqt li tirrikonoxxi restrizzjonijiet le\u0121ittimi tas-sigurt\u00e0 u s-sikurezza. <\/p>\n

Im\u0121ieba Emer\u0121enti u l-Kwistjoni tal-Intelli\u0121enza \u0120enerali<\/h2>\n

Hekk kif il-mudelli lingwisti\u010bi ji\u017cdiedu fl-iskala, ir-ri\u010berkaturi rrappurtaw l-hekk imsej\u0127a abbiltajiet emer\u0121enti: im\u0121ieba li hija dg\u0127ajfa jew assenti f’mudelli i\u017cg\u0127ar i\u017cda ssir notevolment aktar b’sa\u0127\u0127itha f’mudelli akbar. Xog\u0127ol ippubblikat jiddeskrivi dan b\u0127ala effett ta’ evalwazzjoni fejn il-prestazzjoni tidher li tinbidel sew ladarba mudell jaqsam \u010berti limiti ta’ skala u ta\u0127ri\u0121. Il-fenomenu huwa dibattut, i\u017cda jibqa’ influwenti fil-mod kif l-organizzazzjonijiet ibassru l-kapa\u010bit\u00e0 futura. <\/p>\n

Din id-diskussjoni tikkonnettja mal-kun\u010bett tal-intelli\u0121enza \u0121enerali artifi\u010bjali (AGI), \u0121eneralment definita b\u0127ala sistema kapa\u010bi twettaq firxa wiesg\u0127a ta’ kompiti intellettwali fil-livell uman jew ‘il fuq minnu. Jekk l-appro\u010b\u010bi attwali jistg\u0127ux jil\u0127qu l-AGI permezz tal-iskalar biss hija kwistjoni miftu\u0127a. Xi ri\u010berkaturi jargumentaw li l-kompetenza \u0121enerali tista’ to\u0127ro\u0121 minn sistemi ta’ ppro\u010bessar tal-informazzjoni kapa\u010bi bi\u017c\u017cejjed, filwaqt li o\u0127rajn isostnu li l-mudelli tal-lum g\u0127adhom limitati mid-dipendenza tag\u0127hom fuq id-distribuzzjonijiet tat-ta\u0127ri\u0121 u min-nuqqas ta’ fehim ibba\u017cat tag\u0127hom. <\/p>\n

Implikazzjonijiet So\u010bjetali u ta’ Politika<\/h2>\n

Xog\u0127ol, Produttivit\u00e0, u Rwoli \u0120odda<\/h3>\n

L-IA \u0121enerattiva tista’ awtomatizza jew ta\u010b\u010bellera kompiti konjittivi ta’ rutina b\u0127all-abbozzar, il-klassifikazzjoni, it-traskrizzjoni, u l-anali\u017ci tal-ewwel pass. Dan x’aktarx li se jfassal mill-\u0121did il-flussi tax-xog\u0127ol klerikali u professjonali aktar milli jeliminahom b’mod uniformi. B’mod parallel, qed jimmaturaw rwoli \u0121odda, inklu\u017ci operazzjonijiet ta’ tag\u0127lim awtomatiku, sigurt\u00e0 tal-IA, evalwazzjoni tal-mudelli, u governanza tal-IA. L-organizzazzjonijiet li jadottaw dawn l-g\u0127odod b’mod effettiv \u0121eneralment iqabbluhom ma’ responsabbilt\u00e0 \u010bara, sorveljanza umana, u ta\u0127ri\u0121 tal-persunal. <\/p>\n

Xjenza u Kura tas-Sa\u0127\u0127a<\/h3>\n

It-tag\u0127lim awtomatiku di\u0121\u00e0 jintu\u017ca f’oqsma ta’ ri\u010berka b\u0127all-imma\u0121ini medika, l-iskoperta ta’ drogi, u l-bijolo\u0121ija komputazzjonali. F’\u0127afna ka\u017cijiet, dawn is-sistemi ja\u0121ixxu b\u0127ala g\u0127odod ta’ appo\u0121\u0121 g\u0127ad-de\u010bi\u017cjonijiet li jg\u0127inu lix-xjenzati u lill-klini\u010bisti jag\u0127mlu triage tal-informazzjoni jew ji\u0121\u0121eneraw ipote\u017cijiet. L-aqwa ri\u017cultati tipikament ji\u0121u mill-kombinazzjoni tal-outputs tal-AI mal-g\u0127arfien espert tad-dominju, dejta ta’ kwalit\u00e0 g\u0127olja, u validazzjoni robusta. <\/p>\n

Edukazzjoni u Valutazzjoni<\/h3>\n

It-tag\u0127lim privat u l-feedback assistiti mill-AI jistg\u0127u jappo\u0121\u0121jaw tag\u0127lim personalizzat, partikolarment g\u0127all-abbozzar u l-prattika tal-lingwa. Fl-istess \u0127in, l-a\u010b\u010bess fa\u010bli g\u0127al test \u0121enerattiv jisfida l-mudelli tradizzjonali ta’ valutazzjoni. L-istituzzjonijiet edukattivi x’aktarx se jkollhom b\u017conn jadattaw billi jenfasizzaw ix-xog\u0127ol fil-klassi, il-valutazzjoni orali, l-evalwazzjoni bba\u017cata fuq il-pro\u010bess, u struzzjoni espli\u010bita dwar kif tu\u017ca l-AI b’mod responsabbli. <\/p>\n

Kreattivit\u00e0 u Produzzjoni Kulturali<\/h3>\n

Sistemi \u0121enerattivi jistg\u0127u jipprodu\u010bu test, imma\u0121ni, u mu\u017cika malajr, u \u0127afna kreaturi ju\u017cawhom b\u0127ala g\u0127odod ta’ ideazzjoni. G\u0127ad hemm t\u0127assib dwar l-ispostament tax-xog\u0127ol, l-attribuzzjoni, u l-u\u017cu ta’ materjal ta’ ta\u0127ri\u0121 bid-drittijiet tal-awtur. Tikkettar \u010bar, mudelli ta’ li\u010benzjar, u tra\u010b\u010bar tal-provenjenza huma risposti emer\u0121enti, i\u017cda n-normi g\u0127adhom qed jevolvu. <\/p>\n

Governanza, Sigurt\u00e0, u Fidu\u010bja<\/h3>\n

\"Ng\u0127ixu<\/a><\/p>\n

Hekk kif dawn is-sistemi jsiru integrati fis-servizzi pubbli\u010bi u fil-flussi tax-xog\u0127ol tal-intrapri\u017ci, il-governanza ssir rekwi\u017cit ewlieni. Kwistjonijiet ewlenin jinkludu l-protezzjoni tad-dejta, is-sigurt\u00e0 tal-pront u l-outputs, il-pre\u0121udizzju u d-diskriminazzjoni, l-ispjegabbilt\u00e0 fejn id-de\u010bi\u017cjonijiet jaffettwaw id-drittijiet, u r-re\u017ciljenza kontra l-informazzjoni \u017cbaljata. Ir-regolaturi u l-korpi tal-istandards qed jirrispondu b’oqfsa li jenfasizzaw il-\u0121estjoni tar-riskju, it-trasparenza, is-sorveljanza umana, u r-responsabbilt\u00e0 tul i\u010b-\u010biklu tal-\u0127ajja tal-IA. <\/p>\n

Konklu\u017cjoni<\/h2>\n

Mudelli lingwisti\u010bi kbar huma l-a\u0127jar meqjusa b\u0127ala magni statisti\u010bi qawwija g\u0127al-lingwa li \u0121ew iffurmati mill-iskala, id-dejta, u l-irfinar, u mbag\u0127ad ippakkjati fi prodotti li jinfluwenzaw ix-xog\u0127ol ta’ kuljum. Ir-ri\u017cultati tag\u0127hom jistg\u0127u jkunu ta’ g\u0127ajnuna, i\u017cda jistg\u0127u wkoll ikunu \u017cbaljati b’kunfidenza, pre\u0121udikati mid-dejta tat-ta\u0127ri\u0121, jew applikati \u0127a\u017cin f’kuntesti ta’ importanza kbira. Fehim ba\u017ciku ta’ kif dawn il-mudelli huma m\u0127arr\u0121a u kif ji\u0121\u0121eneraw it-test jg\u0127in lill-utenti jevalwaw ir-risposti b’mod aktar kritiku u ju\u017caw l-g\u0127odod b’mod aktar responsabbli. <\/p>\n

Il-fa\u017ci li jmiss tal-IA se tkun iffurmata kemm mill-governanza, l-g\u0127a\u017cliet istituzzjonali, u l-aspettattivi pubbli\u010bi kif ukoll mill-progress tekniku. Jekk is-so\u010bjetajiet iridu l-benefi\u010b\u010bji ta\u2019 dawn is-sistemi ming\u0127ajr riskji ina\u010b\u010bettabbli, se jkollhom b\u017conn utenti infurmati, organizzazzjonijiet kapa\u010bi, u politiki li jippremjaw it-trasparenza u r-responsabbilt\u00e0. Il-fehim, aktar milli l-adozzjoni passiva, huwa l-aktar punt tat-tluq prattiku. <\/p>\n

Referenzi u Aktar Qari<\/h2>\n

Is-sorsi li \u0121ejjin jipprovdu punti tad-d\u0127ul a\u010b\u010bessibbli g\u0127all-pedamenti tekni\u010bi u d-dibattiti dwar il-governanza li jdawru mudelli lingwisti\u010bi kbar u sistemi relatati.<\/p>\n

    \n
  • Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, AN, Kaiser, L., u Polosukhin, I. (2017). Attenzjoni Hija Kulma g\u0127andek b\u017conn<\/em> . arXiv:1706.03762. <\/li>\n
  • Brown, TB, Mann, B., Ryder, N., Subbiah, M., Kaplan, JD, Dhariwal, P., u o\u0127rajn. (2020). Il-Mudelli Lingwisti\u010bi huma Studenti b’Ftit Skadenza<\/em> . arXiv:2005.14165. <\/li>\n
  • Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., u o\u0127rajn. (2020). \u0120enerazzjoni mi\u017cjuda bl-Irkupru g\u0127al Kompiti NLP Intensivi ta’ G\u0127arfien<\/em> . arXiv:2005.11401. <\/li>\n
  • Bommasani, R., Hudson, DA, Adeli, E., Altman, R., Arora, S., u o\u0127rajn. (2021). Dwar l-Opportunitajiet u r-Riskji tal-Mudelli ta’ Fondazzjoni<\/em> . arXiv:2108.07258. <\/li>\n
  • Wei, J., Tay, Y., Bommasani, R., Raffel, C., Zoph, B., u o\u0127rajn. (2022). Abbiltajiet Emer\u0121enti ta’ Mudelli ta’ Lingwi Kbar<\/em> . arXiv:2206.07682. <\/li>\n
  • Bender, EM, Gebru, T., McMillan-Major, A., u Shmitchell, S. (2021). Dwar il-Perikli tal-Pappagalli Stokasti\u010bi: Jistg\u0127u l-Mudelli Lingwisti\u010bi jkunu Kbar Wisq?<\/em> Atti tal-Konferenza tal-ACM dwar il-\u0120ustizzja, ir-Responsabbilt\u00e0, u t-Trasparenza (FAccT).<\/li>\n
  • Tabassi, E. (2023). Qafas g\u0127all-\u0120estjoni tar-Riskju tal-Intelli\u0121enza Artifi\u010bjali (AI RMF 1.0)<\/em> . Istitut Nazzjonali tal-Istandards u t-Teknolo\u0121ija (NIST AI 100-1). <\/li>\n
  • OECD (2019). Prin\u010bipji tal-OECD dwar l-Intelli\u0121enza Artifi\u010bjali<\/em> . Organizzazzjoni g\u0127all-Kooperazzjoni u l-I\u017cvilupp Ekonomi\u010bi. <\/li>\n<\/ul>\n

    \u010aa\u0127da ta’ Responsabbilt\u00e0 Editorjali<\/em><\/h2>\n

    Dan id-dokument t\u0127ejja bl-g\u0127ajnuna ta’ g\u0127odod \u0121enerattivi tal-IA g\u0127al appo\u0121\u0121 editorjali (e\u017c., raffinament tal-lingwa u su\u0121\u0121erimenti g\u0127all-abbozzar). Kwalunkwe su\u0121\u0121eriment i\u0121\u0121enerat mill-IA \u0121ie rivedut, editjat, u vvalidat mill-awturi, li jibqg\u0127u kompletament responsabbli g\u0127all-kontenut finali. Fejn rilevanti, intu\u017caw ukoll g\u0127odod tal-IA biex ji\u017cviluppaw kun\u010betti illustrattivi u su\u0121\u0121erimenti g\u0127all-\u0121enerazzjoni tal-imma\u0121ni; id-de\u010bi\u017cjonijiet finali dwar l-g\u0127a\u017cla u l-inklu\u017cjoni ttie\u0127du mill-awturi u l-pubblikatur. <\/em><\/p>\n

    Dan l-artiklu huwa bba\u017cat fuq informazzjoni mi\u0121bura minn diversi sorsi online u ma\u0127sub biss g\u0127al skopijiet informattivi. Ma jikkostitwixxix parir jew rakkomandazzjonijiet professjonali. Filwaqt li l-informazzjoni pprovduta hija pre\u010bi\u017ca u vera sa fejn nafu a\u0127na, a\u0127na ma nag\u0127mlu l-ebda rappre\u017centazzjoni jew garanzija ta\u2019 kwalunkwe tip, espressa jew implikata, rigward il-kompletezza, l-e\u017cattezza, il-fidu\u010bja, l-adegwatezza, jew id-disponibbilta\u2019 tas-su\u0121\u0121etti koperti. A\u0127na nav\u017caw bil-qawwa lill-qarrejja biex iwettqu aktar ri\u010berka u jikkonsultaw professjonisti jew esperti xierqa qabel ma jie\u0127du kwalunkwe de\u010bi\u017cjoni bba\u017cata fuq l-informazzjoni ppre\u017centata f\u2019dan l-Artikolu. <\/em><\/p>\n

    A\u0127na ni\u010b\u0127du kwalunkwe responsabbilt\u00e0 g\u0127al telf jew danni m\u0121arrba bba\u017cati fuq l-informazzjoni pprovduta hawnhekk.<\/em><\/p>\n","protected":false},"author":7,"template":"","class_list":["post-98121","article","type-article","status-publish","has-post-thumbnail","hentry"],"yoast_head":"\nNg\u0127ixu b'Mudelli Lingwisti\u010bi Kbar - Malta Digital Innovation Authority<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/\" \/>\n<meta property=\"og:locale\" content=\"mt_MT\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Ng\u0127ixu b'Mudelli Lingwisti\u010bi Kbar - Malta Digital Innovation Authority\" \/>\n<meta property=\"og:description\" content=\"Astratt Mudelli lingwisti\u010bi kbar m’g\u0127adhomx limitati g\u0127al-laboratorji tar-ri\u010berka. Huma integrati f’applikazzjonijiet tal-konsumatur u pjattaformi tax-xog\u0127ol, u issa jinfluwenzaw kif in-nies […]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/\" \/>\n<meta property=\"og:site_name\" content=\"Malta Digital Innovation Authority\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/ESkillsMalta\" \/>\n<meta property=\"og:image\" content=\"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"944\" \/>\n\t<meta property=\"og:image:height\" content=\"631\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:site\" content=\"@eSkills_Malta\" \/>\n<meta name=\"twitter:label1\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"10 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/\",\"url\":\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/\",\"name\":\"Ng\u0127ixu b'Mudelli Lingwisti\u010bi Kbar - Malta Digital Innovation Authority\",\"isPartOf\":{\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg\",\"datePublished\":\"2026-05-13T07:03:59+00:00\",\"breadcrumb\":{\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#breadcrumb\"},\"inLanguage\":\"mt-MT\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"mt-MT\",\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#primaryimage\",\"url\":\"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg\",\"contentUrl\":\"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg\",\"width\":944,\"height\":631,\"caption\":\"Living with Large Language Models\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/digitalskills.mdia.gov.mt\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Article\",\"item\":\"https:\/\/digitalskills.mdia.gov.mt\/article\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"Ng\u0127ixu b’Mudelli Lingwisti\u010bi Kbar\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/#website\",\"url\":\"https:\/\/digitalskills.mdia.gov.mt\/\",\"name\":\"Malta Digital Innovation Authority\",\"description\":\"LISP\",\"publisher\":{\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/digitalskills.mdia.gov.mt\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"mt-MT\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/#organization\",\"name\":\"eSkills Malta Foundation\",\"url\":\"https:\/\/digitalskills.mdia.gov.mt\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"mt-MT\",\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/#\/schema\/logo\/image\/\",\"url\":\"\/wp-content\/uploads\/2023\/04\/eskills-Foundation-logo.jpg\",\"contentUrl\":\"\/wp-content\/uploads\/2023\/04\/eskills-Foundation-logo.jpg\",\"width\":500,\"height\":217,\"caption\":\"eSkills Malta Foundation\"},\"image\":{\"@id\":\"https:\/\/digitalskills.mdia.gov.mt\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/www.facebook.com\/ESkillsMalta\",\"https:\/\/x.com\/eSkills_Malta\"]}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Ng\u0127ixu b'Mudelli Lingwisti\u010bi Kbar - Malta Digital Innovation Authority","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/","og_locale":"mt_MT","og_type":"article","og_title":"Ng\u0127ixu b'Mudelli Lingwisti\u010bi Kbar - Malta Digital Innovation Authority","og_description":"Astratt Mudelli lingwisti\u010bi kbar m’g\u0127adhomx limitati g\u0127al-laboratorji tar-ri\u010berka. Huma integrati f’applikazzjonijiet tal-konsumatur u pjattaformi tax-xog\u0127ol, u issa jinfluwenzaw kif in-nies […]","og_url":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/","og_site_name":"Malta Digital Innovation Authority","article_publisher":"https:\/\/www.facebook.com\/ESkillsMalta","og_image":[{"width":944,"height":631,"url":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","type":"image\/jpeg"}],"twitter_card":"summary_large_image","twitter_site":"@eSkills_Malta","twitter_misc":{"Est. reading time":"10 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/","url":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/","name":"Ng\u0127ixu b'Mudelli Lingwisti\u010bi Kbar - Malta Digital Innovation Authority","isPartOf":{"@id":"https:\/\/digitalskills.mdia.gov.mt\/#website"},"primaryImageOfPage":{"@id":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#primaryimage"},"image":{"@id":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#primaryimage"},"thumbnailUrl":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","datePublished":"2026-05-13T07:03:59+00:00","breadcrumb":{"@id":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#breadcrumb"},"inLanguage":"mt-MT","potentialAction":[{"@type":"ReadAction","target":["https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/"]}]},{"@type":"ImageObject","inLanguage":"mt-MT","@id":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#primaryimage","url":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","contentUrl":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","width":944,"height":631,"caption":"Living with Large Language Models"},{"@type":"BreadcrumbList","@id":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/digitalskills.mdia.gov.mt\/"},{"@type":"ListItem","position":2,"name":"Article","item":"https:\/\/digitalskills.mdia.gov.mt\/article\/"},{"@type":"ListItem","position":3,"name":"Ng\u0127ixu b’Mudelli Lingwisti\u010bi Kbar"}]},{"@type":"WebSite","@id":"https:\/\/digitalskills.mdia.gov.mt\/#website","url":"https:\/\/digitalskills.mdia.gov.mt\/","name":"Malta Digital Innovation Authority","description":"LISP","publisher":{"@id":"https:\/\/digitalskills.mdia.gov.mt\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/digitalskills.mdia.gov.mt\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"mt-MT"},{"@type":"Organization","@id":"https:\/\/digitalskills.mdia.gov.mt\/#organization","name":"eSkills Malta Foundation","url":"https:\/\/digitalskills.mdia.gov.mt\/","logo":{"@type":"ImageObject","inLanguage":"mt-MT","@id":"https:\/\/digitalskills.mdia.gov.mt\/#\/schema\/logo\/image\/","url":"\/wp-content\/uploads\/2023\/04\/eskills-Foundation-logo.jpg","contentUrl":"\/wp-content\/uploads\/2023\/04\/eskills-Foundation-logo.jpg","width":500,"height":217,"caption":"eSkills Malta Foundation"},"image":{"@id":"https:\/\/digitalskills.mdia.gov.mt\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/ESkillsMalta","https:\/\/x.com\/eSkills_Malta"]}]}},"acf":{"digital_skill_level":["basic","intermediate","advanced"],"digital_technology":["http:\/\/data.europa.eu\/uxp\/3030","http:\/\/data.europa.eu\/uxp\/437655"],"geographic_scope":[{"value":"http:\/\/publications.europa.eu\/resource\/authority\/country\/EUR","label":"European Union"},{"value":"http:\/\/publications.europa.eu\/resource\/authority\/country\/MLT","label":"Malta"}],"initiative_type":"local_initiative","image":{"ID":98117,"id":98117,"title":"Living-with-Large-Language-Models","filename":"Living-with-Large-Language-Models.jpg","filesize":443870,"url":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","link":"https:\/\/digitalskills.mdia.gov.mt\/mt\/article\/nghixu-bmudelli-lingwistici-kbar\/living-with-large-language-models-2\/","alt":"Living with Large Language Models","author":"7","description":"","caption":"","name":"living-with-large-language-models-2","status":"inherit","uploaded_to":98121,"date":"2026-05-13 07:00:21","modified":"2026-05-13 07:06:49","menu_order":0,"mime_type":"image\/jpeg","type":"image","subtype":"jpeg","icon":"https:\/\/digitalskills.mdia.gov.mt\/wp-includes\/images\/media\/default.png","width":944,"height":631,"sizes":{"thumbnail":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models-150x150.jpg","thumbnail-width":150,"thumbnail-height":150,"medium":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","medium-width":944,"medium-height":631,"medium_large":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models-768x513.jpg","medium_large-width":768,"medium_large-height":513,"large":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","large-width":944,"large-height":631,"1536x1536":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","1536x1536-width":944,"1536x1536-height":631,"2048x2048":"https:\/\/digitalskills.mdia.gov.mt\/wp-content\/uploads\/2026\/05\/Living-with-Large-Language-Models.jpg","2048x2048-width":944,"2048x2048-height":631}},"publish_in_core_platform":true,"author":"Bernard Montebello, Trevor Sammut","link":"","link_text":"","link_type":"organization_url"},"_links":{"self":[{"href":"https:\/\/digitalskills.mdia.gov.mt\/mt\/wp-json\/wp\/v2\/articles\/98121","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/digitalskills.mdia.gov.mt\/mt\/wp-json\/wp\/v2\/articles"}],"about":[{"href":"https:\/\/digitalskills.mdia.gov.mt\/mt\/wp-json\/wp\/v2\/types\/article"}],"author":[{"embeddable":true,"href":"https:\/\/digitalskills.mdia.gov.mt\/mt\/wp-json\/wp\/v2\/users\/7"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/digitalskills.mdia.gov.mt\/mt\/wp-json\/wp\/v2\/media\/98117"}],"wp:attachment":[{"href":"https:\/\/digitalskills.mdia.gov.mt\/mt\/wp-json\/wp\/v2\/media?parent=98121"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}