Cabanga lokhu: uthisha wesiNgisi webanga eliphakathi/esikoleni samabanga aphezulu uhlezi etafuleni lakhe ngoMgqibelo kusihlwa, ikhofi selipholile eceleni kwakhe, egxile ekubhekeni inqwaba yezindatshana zabafundi. Okuthile kumisa umoya wakhe. Amagama asetshenzisiwe acwengisisiwe, izimpikiswano ziqinile, ukuxhumanisa imibono (transitions) kukushelelayo, kodwa kunomuzwa wokuthi kuke kwenzeka okungahambi kahle. Akusakhulumi njengomfundi owake waphazima wahluleka ukukhuluma ngokuxoxa ekilasini ezinsukwini ezimbalwa ngaphambili. Uyifaka ekuhloleni okulula kokukopela (basic plagiarism checker), kubuye kuhlanzekile. Uyazama nomtshina we-AI wamahhala (free AI detector), umphumela ungacaci. Uhlala nomuzwa wesisu, lingekho ubufakazi, futhi akukho indlela ecacile yokuthi enzeni elandelayo.
Lesi sigameko siyenzeka emakilasini emhlabeni wonke. Kusukela amathuluzi okubhala e-AI afana neChatGPT, Gemini, neClaude asetholakala kabanzi futhi mahhala, othisha bazithole besemgqeni ongabekezeleleki: ukugcina amazinga obunembi bezemfundo (academic integrity standards) okwakungakhelwanga umhlaba lapho umfundi engakhiqiza khona indatshana ephelele, evakala njengowokuqala, ngaphansi kwemizuzwana engamashumi amathathu. Umbuzo awusasekho ukuthi i-AI iyashintsha imfundo yini. Iyashintsha. Umbuzo wangempela ukuthi othisha kumele benzeni ngayo.
Imithetho Yudala Ayisasebenzi
Sekuyiminyaka eminingi, imithetho yobunembi bezemfundo yakhiwa phezu kwesisekelo esilula: uma umfundi ethumele umsebenzi ongowakhe, umtshina wokukopela uzowubamba ngokuwuhlanganisa nombhalo osuvele ukhona kweminye imithombo (database) ekhona. Lawa mathuluzi abe yinto ejwayelekile ezikoleni nasezikhungweni zemfundo ephakeme ngoba ukukopela kwe-AI, ngendlela ekhona manje, kwakungakabi khona.
Manje la mathuluzi sehlukumezeka kakhulu ekubambeni okuqukethwe okukhiqizwe yi-AI. Uma umfundi ekopela umbhalo kuwebhusayithi noma ephepheni elishicilelwe, lowo mbhalo usuvele ukhona kwenye indawo futhi ungafakwa uphawu. Kodwa uma umfundi ecela i-AI ukuthi ibhale indatshana, okuphumayo kuvela okusha. Akukho dokhumenti yokuqala okufanele iqhathaniswe nayo. Ama-plagiarism checkers ajwayelekile awakhiwanga ukuthi athole ukubhala okwenziwa yi-AI, futhi akunakwenzeka ukuthi noma yiluphi “ukulungisa” kula masistimu amadala luwenze afanele inkinga othisha ababhekene nayo manje.
Ukuze izinto zibe nzima kakhulu, okuqukethwe okukhiqizwe yi-AI manje kungenziwa kuhumushwe phakathi kwezilimi bese kuthunyelwa kungabonakali. Umfundi angase acele i-AI ukuthi ibhale indatshana ngolunye ulimi, bese eyidlulisela endaweni yokuhumusha ngaphambi kokuyithumela. Ama-plagiarism checkers ajwayelekile ahlola ulimi olulodwa kuphela uzowushaya indiva lo mehluko ngokuphelele, yingakho ukutholwa kokukopela okuhumushiwe phakathi kwezilimi (cross-language translated plagiarism detection) sekuyingxenye ebalulekile yanoma iyiphi i-toolkit ehlonipha kakhulu ubuqiniso bezemfundo.
Umsamo phakathi kwenqubomgomo yesikhungo kanye neqiniso eliseklasini awukakaze ube banzi kangaka. Izikole eziningi zisasebenzisa ama-handbook obunembi bezemfundo abhalwe iminyaka noma ngisho neminyaka engamashumi eminyaka edlule. Imisho efana nokuthi “ukuthumela umsebenzi ongowakho” iba nzima ngokomqondo uma umfundi eqinisweni ebethe ithegi (typed the prompt), wayibuyekeza imiphumela, futhi mhlawumbe wenze izinguquko ezincane endleleni. Imithetho ayikashintshi ngokuhambisana nesikhathi, kanti othisha abaphoqelelayo baphendukela ekutolikeni izindawo ezingwevu (gray areas) bona bodwa, bengenaseluleko esifanele noma ukwesekwa.
I-Dilemma Yothisha
Ngaphezu kwenkinga yenqubomgomo, kukhona inkinga ejulile ebantwini. Othisha nabeluleki (professors) babekwa endaweni engakhululekile yokuba sengathi bangabaseshi, futhi phezulu kakhulu izindleko ngakolunye uhlangothi nakwelinye.
Ukusola umfundi ngokusebenzisa i-AI ngaphandle kobufakazi obucacile kuyindaba enkulu. Kungalimaza amarekhodi emfundo yomfundi, kuqinise ukungezwani phakathi kothisha nomfundi, futhi kwezinye izimo kungaholela ezinqubweni ezisemthethweni zokujezisa. Kodwa futhi, ukuhlala uthule uma kusolwa kakhulu ukuthi kukhona ukukopela nge-AI kuzizwa sengathi ukukhashela konke okufanele imele ubuqiniso bezemfundo. Othisha banamathele phakathi kokuvikela abafundi ekusolweni okungenabulungisa nokuvikela inani lomsebenzi oqotho.
Lokhu kungaqiniseki kuhudula umthwalo obambekayo. Othisha abaningi babika ukuthi bazizwa becindezelekile, bengasizi, futhi bengatholi ukwesekwa lapho behamba nalezi zimo. Isisindo esingokomzwelo sokungakwazi ukuthembela emsebenzini othunyelwe, sokungabaze yonke indima ebhalwe kahle, nokuzibuza ukuthi umfundi uthole ibanga lakhe ngenxa yokuzikhandla noma walikhipha ngemshini, kunciphisa kancane kancane injabulo yokufundisa kubo abaningi. Ukuthembana, okwakungumgogodla othulile wekhela lekilasi, manje sekungaphansi kokucindezeleka ngendlela enzima ukuyilungisa.
Othisha abakudingayo akusona nje isicelo sokuthola (detection tool) kuphela, kodwa i-workflow ephelele ebabasizayo ukuthi babone izinkinga ezingaba khona, baqonde ukuthi ziyini lezo zinkinga, bese benza okuthile kuzo ngokuzethemba. Lokho kuphakeme kakhulu kunalokho amathuluzi amaningi akhona aklanyelwe ukukuhlangabezana.
Kungani Amathuluzi Okuhlola Okujwayelekile (Generic) We-AI Enganele
Njengokuphendula ekukhuphukeni kokuqukethwe okukhiqizwe i-AI ezimeni zemfundo, kwangena emakethe uchungechunge lwamathuluzi okutotoshela (AI detector tools), kuthembisa ukuxazulula inkinga. Amathuluzi athi awakwazi ukuthola ukubhala kwe-AI ngokunemba okukhulu athandwa ngokushesha, kodwa iqiniso selivele liyinkimbinkimbi kakhulu.
Inkinga enkulu ngamathuluzi amaningi okuhlola i-AI ukuthi awathembekile. Ucwaningo nokuhlola empilweni yangempela kukhombisile ngokungaguquki ukuthi la mathuluzi akhiqiza amazinga aphezulu okuthi kube “amaphutha amabili” (kokubili false positives kanye false negatives). I-false positive isho ukuthi indatshana ebhalwe ngumuntu ibekwa uphawu njengokwakhiqizwe yi-AI, okungaholela ekutheni umfundi omsulwa asolwe ngokukhohlisa. I-false negative isho ukuthi okuqukethwe okwenziwe i-AI kwangempela kudlula kungabanjwanga. Zombili lezi zimo azisizi othisha noma abafundi.
Okukwenza kube kubi nakakhulu, amaningi alawa mathuluzi asebenza ngolimi lwesiNgisi kuphela. Ezikoleni nasezikhungweni ezinemiphakathi yezilimi eziningi (multilingual), lokhu kuwukulinganiselwa okukhulu. Abafundi ababhala ngeSpanish, Filipino, French, Arabic, noma ezinye izilimi eziningi bahlala bengabonakali kulawa mathuluzi okuthola, ngoba la mathuluzi akhelwe cabangela ulimi olulodwa kuphela.
Amathuluzi okubhala we-AI nawo ayashintsha ngokushesha futhi manje angacelwa ukuthi abhale ngendlela engajwayelekile, engaphelele, enokuzwakala “komuntu” okuzodinga ngokukhethekile ukugwema ukutholwa. Abafundi bathole ukuthi uma ubacela i-AI ukuthi ibhale ngezici ezihlosiwe eziphazamisa kancane (intentional quirks) noma ngendlela ekhulumisanayo, kungakhohlisa amathuluzi amaningi okuhlola indatshana ye-AI. Ubuchwepheshe obusetshenziselwa ukuthola ukubhala kwe-AI buhlala buyisinyathelo esingaphambili kancane ubuchwepheshe obukhiqiza i-AI, yingakho ukuhlaziywa kwesigaba ngesigaba (sentence-level breakdown) kunokuba kube isikolo esisodwa kuphela (single overall score) kubalulekile kubo othisha abafuna ukwazi ngokucacile ukuthi i-AI yasetshenziswa kuphi nokuthi yasetshenziswa kanjani kulo mbhalo.
Okubonakala Njengethuluzi Elithembekile Lobunembi Bezifundo
Akuwona wonke amathuluzi okuthola ukukopela (plagiarism) nokuhlola i-AI afana, futhi umehluko ubaluleke kakhulu uma kunqunywa izinqumo zemfundo. Ithuluzi eliwusizo ngempela kubo othisha kumele lenze izinto eziningana kahle ngesikhathi esisodwa.
Okokuqala, kumele libe nekhono lokusebenza ngezilimi eziningi (multilingual). Izikhungo zemfundo emhlabeni wonke zisebenza ngezilimi ezingaphezu kwamashumi amaningi, futhi ithuluzi elibamba kuphela ukukopela kwe-AI ngesiNgisi alikona ngempela elisiza umphakathi wemfundo waphesheya komhlaba. I-Plag.ai’s AI detector isekela izilimi ezingaphezu kuka-50 zokuthola i-AI kanye nezilimi ezingaphezu kuka-100 zokuhlola ukukopela, okusho ukuthi othisha ePhilippines, kuwo wonke amazwe aseYurophu, Latin America, nase-Asia bangathembela endaweni efanayo ngaphandle kokulahlekelwa ukunemba ngenxa yolimi lombhalo othunyelweyo.
Okwesibili, kumele kuya phambili kunesikolo esisodwa. Ithuluzi elitshela othisha ukuthi umbhalo “ufana ngama-74%” ngaphandle kokukhombisa ukuthi yimaphi amasentensi athole uphawu (flagged), alisebenzi kakhulu. Othisha abakudingayo ukuhlaziywa kwamasentensi (sentence-level breakdown) okugqamisa ngokucacile ukuthi yiziphi izingxenye zomsebenzi othunyelweyo ezingase zibe yi-AI generated noma zikopeliwe, kanye nama-link emithombo lapho kwatholakala ukufana khona. Leli zinga lemininingwane lenza kube lula ukuba kube nengxoxo enolwazi, esekelwe ebufakazini nomfundi kunokwenza isinqumo ngokulinganisa amathuba nje angacaci.
Okwesithathu, kumele libambe ukukopela okuhumushiwe. I-Plag.ai inikeza ukutholwa kokukopela okuhumushiwe phakathi kwezilimi (cross-language translated plagiarism detection), okuyisici esikhethekile esibonisa uma okuqukethwe kuhumushiwe kusuka kwenye ulimi ngaphambi kokuthunyelwa. Lokhu kuvala enye yezimbobo ezinkulu kakhulu ekutholeni ukukopela okujwayelekile futhi kunika othisha isithombe esiphelele kakhulu sobuqiniso bomsebenzi wokuqala wombhalo.
Okwesine, kumele kukhiqize umbiko ongadawunilodwa (downloadable) futhi wabelwane. Uma uthisha ethola inkinga engaba khona yobunembi, udinga ukuyibhalisa. I-Plag.ai ikhiqiza umbiko we-PDF ongadawunilodwa we-originality (originality report) ongabelwa nabaphathi, abafundi, noma amaqembu abheka ubuqiniso bezemfundo (academic integrity committees), unikeze indlela ecacile yamarekhodi (paper trail) evikela bobabili uthisha nomfundi ngesikhathi sokubuyekezwa.
Okokugcina, okubaluleke kakhulu ezikhungweni zemfundo, kumele kuvikele ubumfihlo (privacy). Esinye sezinto ezikhathaza kakhulu kubo othisha nabafundi uma bethumela imibhalo emathuluzini angabantu (third-party tools) wubungozi bokuthi leyo mibhalo ingangezwa kusizindalwazi sokuqhathanisa noma yabelwe ezinye izikhungo. I-Plag.ai isebenza ngomgomo oqinile wokubeka ubumfihlo kuqala: imibhalo ayisoze yabelwa izikhungo, ayisoze yengezwe kusizindalwazi sokuqhathanisa, futhi ayisoze isatshalaliswe kubantu besithathu. Okungokwakho kuhlala kungokwakho.
Lokho Othisha Abazama Ukukwenza Eklasini
Ngenxa yamathuluzi anganele kanye nezinqubomgomo ezindala, othisha abaningi sebeqale ukucabanga kabusha indlela yabo kusukela phansi. Kunokuba bazame ukubamba ukusetshenziswa kwe-AI sekwephuzile, abanye baklama imisebenzi (assignments) ngendlela eyenza okuqukethwe okukhiqizwe yi-AI kungabi usizo kakhulu ekuqaleni.
Enye yezindlela ezisebenza kakhulu ethandwa kakhulu wukubuyisela ukuhlolwa kokubhala emuva ekilasini. Imisebenzi yokubhala eyenziwa phansi kokugadwa (under supervision) iyisuse ngokuphelele ithuba lokuthi i-AI ingene. Abanye othisha bayahlanganisa lokhu nokuzivikela ngomlomo (oral defenses), lapho abafundi kumele bachaze futhi bandise ngomlomo umsebenzi obhaliwe abawuthumele. Uma umfundi engakwazi ukukhuluma ngemibono esendatshaneni yakhe, igebe liyabonakala ngaphandle kokuthi kudingeke noma iyiphi i-AI detector.
Abanye futhi bakhetha imiyalelo yokusebenza ecaciswe kakhulu, ejulile, nephethwe yindlela yomuntu (hyper-specific, deeply personal assignment prompts). Uma ucela abafundi ukuba babhale ngemicimbi ethile yendawo, ngesipiliyoni somuntu siqu, noma isihloko esincane kakhulu esidinga ulwazi lwangempela (firsthand knowledge), kuba nzima kakhulu ku-AI ukukhiqiza okuzwakalayo kuqinisile. Amathuluzi e-AI asebenza kangcono uma enikezwa imiyalelo ebanzi, ejwayelekile. Uma umsebenzi uba bucacile futhi ube nenhliziyo yomuntu, i-AI iba nenzuzo encane.
Ukubeka amaphuzu ngokubheka inqubo (process-based grading) kungenye indlela eyanda. Esikhundleni sokuhlola kuphela umbhalo wokugcina othunyelweyo, othisha manje sebecele abafundi ukuthi bathumele amanothi okucabanga (brainstorming notes), okusalungiswa okuningana (multiple drafts), amarekhodi wokubuyekezwa kontanga (peer review records), kanye nemibiko yocwaningo (research logs) kanye nomsebenzi wabo wokugcina. Lolu chungechunge lwamarekhodi (paper trail) lwenza kube nzima kakhulu ukufihla noma ukulingisa inqubo yokufunda ngoba injongo yomsebenzi iguquka isuke ekukhiqizeni umkhiqizo ophucuziwe iye ekukhombiseni ukuthuthuka okusebenzayo kwengqondo (genuine intellectual development) ngokuhamba kwesikhathi.
Kubafundisi abafuna ukusekela abafundi kunokubajezisa nje, amathuluzi afana nesevisi yokususwa kokukopela yePlag.ai (plagiarism removal service) kanye nesevisi yokwenza umbhalo ube “njengomuntu” (expert humanization service) anikeza indlela ehlukile yokwakha phambili. Esikhundleni sokubheka umbhalo osuke watholakala unophawu njengokuphela komgwaqo (dead end), lezi zinsiza zisiza abafundi baqonde ukuthi yini evezwe njengosolisayo nokuthi kumele bayibhale kabusha kanjani ngendlela efanele, okuguqula isigameko esingaba ubuqiniso bezemfundo sibe yithuba langempela lokufunda. Abafundi bangaphinde basebenzise ukuhlola kokukopela okukhululekile ukuze bahlole umsebenzi wabo ngaphambi kokuthumela, okukhuthaza isiko lokuzihlola (self-checking) kanye nobuqiniso (originality) kunokusuka ekugwemeni nokungabaza.
Ingxoxo Enkulu Izikole Okudingeka Ziyibe Nayo
Kuyoba iphutha ukukubeka lokhu njengomsebenzi okumele uthisha ngamunye awuxazulule yedwa. Ukukhula kokuqukethwe okukhiqizwe yi-AI ezimeni zemfundo kuyinselelo yesistimu (systemic challenge) edinga impendulo yesistimu, futhi othisha akufanele bashiywe bazame ukuyiqonda iklasi ngalinye nomsebenzi ngamunye.
Izikole namanyuvesi kufanele zibheke ngokujulile imithetho yazo yobunembi bezemfundo bese beyibuyekeza ukuze ibhekane ne-AI ngokucacile. Lokhu kusho ukuchaza ngokusobala ukuthi yini okuthathwa njengokwamukelekayo nokungamukeleki ekusebenziseni i-AI, ngoba akukona konke ukusetshenziswa kwe-AI okulingana nokukopela kwe-AI. Ukusebenzisa i-AI ukusiza ekuhlakuleleni imibono kuyinto ehlukile ngokuphelele ekuthumeleni umsebenzi oqedwe ngokuphelele yi-AI njengowakho. Imithetho ecacile, enemininingwane (nuanced), isiza kokubili abafundi nabafundisi ukuthi bahambe kulezi zinhlukaniso ngaphandle kokudideka.
Abaphathi nabo banomthwalo wokunikeza othisha ukuqeqeshwa, izinsiza, nezinhlelo zokwesekwa ezisesikhathini. I-Plag.ai iyayiqaphela le nkinga ngokuhlinzeka nge-akhawunti yamahhala kathisha (free educator account) evumela othisha, oprofesa, nabafundisi (lecturers) ukuthi bahlolisise imibhalo engafika ku-20 ngomunye ngenyanga ngaphandle kokukhokha, kanye nethuba lokuthola imibiko eyabiwe ngabafundi ngqo ngepulatifomu. Lokhu kusho ukuthi othisha bangaqala ngaphandle kokuvinjelwa yisabelomali, futhi abafundi bangabelana ngemiSebenzi yokuqinisekisa ubuqiniso (originality reports) yabo othisha njengengxenye yenqubo yokuthumela, kwakhe indlela ecacile nenokusebenzisana ekubhekaneni nobunembi bezemfundo.
Abenzi bencebo (policymakers) emazingeni esifunda nezwe nabo kumele bangene kule ngxoxo. I-AI emfundweni akuyona into engeyona eyomkhakha othile. Iyashintsha lonke uhlaka lokufunda nokuhlola, futhi impendulo ehlukaniswe iklasi ngalinye ngeke yanele. Umhlahlandlela ohlelekile, uxhaso locwaningo ukuze kube nokutholwa okungcono, kanye nokuhlanganiswa okuhle kwamathuluzi athembekile afana nePlag.ai ezinhlelweni zesikhungo konke kuyingxenye yesixazululo esikhulu.
Isiphetho
Ukukhula kwamathuluzi okubhala e-AI akukhombanga nje indlela entsha yokukhohlisa. Kwaphoqa ukucabanga kabusha okuyisisekelo kokuthi imfundo iyenzelwani ngempela. Uma inhloso yomsebenzi wokubhala iwukuthi kukhiqizwe nje idokhumenti ephucuziwe, khona-ke i-AI iye yenza leyo nhloso ibe lula kakhulu ukuyiyikhipha kwabanye (outsource). Kodwa uma inhloso iwukuthuthukisa ukucabanga okujulile (critical thinking), ukuzijwayeza ukuxhumana ngemibono eyinkimbinkimbi, nokukhombisa ukuqonda okungempela, khona-ke i-AI ayikwazi ukuyikhipha lokho, futhi othisha banethuba lokuklama ukuhlolwa okubonisa lezo zinhloso ezijulile.
Impendulo akusona isenzo sokulwa impi engenathemba nobuchwepheshe okuzoqhubeka buba ngcono. Impendulo ukuguquka ngokucabangisisa, ukuxhobisa othisha ngamathuluzi asebenza ngempela, nokwakha amasistimu enza ubuqiniso kube lula ukuyigcina kunokuyigwema. Lokho kusho ukukhetha amathuluzi okuhlola ukukopela ne-AI anemissebenzi yezilimi eziningi (multilingual), anembile (precise), agxile kubumfihlo (privacy-focused), futhi akhelwe izimo zemfundo yesimanje—hhayi ikilasi lesikhathi seminyaka eyishumi edlule.
I-Plag.ai yakhelwe ngqo lokhu engqondweni. Inathemba kubafundi abangaphezu kuka-1.5 million futhi isetshenziswa othisha emhlabeni wonke, ihlanganisa ukuhlola ukukopela (plagiarism checking), ukuhlola i-AI (AI detection), ukutholwa kokukopela okuhumushiwe (translated plagiarism detection), nezinsiza zokwesekwa ezivela kochwepheshe kube yindawo eyodwa esebenza kuwo wonke umphakathi wezemfundo. Noma ngabe unguthisha ozama ukuvikela ubuqiniso bekilasi lakho, noma ungumfundi ofuna ukuthumela ngokuzethemba, i-Plag.ai ikunika amathuluzi okwenza ngendlela efanele.
Ngakho nansi umbuzo ofanele ukuwuhlala ubhekile: esikhundleni sokuzibuza ukuthi sibamba kanjani abafundi abasebenzisa i-AI, kuthiwani uma siqala ukuzibuza ukuthi sakha kanjani isiko lezemfundo lapho kukhuthazwa khona ubuqotho, ukunqotshwa kobuqiniso (originality) kuyazuzisa, futhi amathuluzi afanele enza ubuqiniso kube indlela engahambisani nobunzima (path of least resistance)?
