{"id":21401,"date":"2024-09-26T08:23:00","date_gmt":"2024-09-26T08:23:00","guid":{"rendered":"https:\/\/virtual-routes.org\/?p=21401"},"modified":"2025-01-22T10:09:07","modified_gmt":"2025-01-22T10:09:07","slug":"gestionarea-provocarilor-legate-de-detectarea-ia-in-mediul-academic","status":"publish","type":"post","link":"https:\/\/virtual-routes.org\/ro\/gestionarea-provocarilor-legate-de-detectarea-ia-in-mediul-academic\/","title":{"rendered":"Gestionarea provoc\u0103rilor legate de detectarea IA \u00een mediul academic"},"content":{"rendered":"

Cre\u0219terea num\u0103rului de chatbots AI \u0219i de instrumente de detectare<\/h2>\n

Lansarea ChatGPT \u00een 2022 a generat interes la nivel mondial pentru inteligen\u021ba artificial\u0103 (AI) \u0219i a condus la o utilizare pe scar\u0103 larg\u0103 a chatbots AI, inclusiv de c\u0103tre studen\u021bi. \u00cen urma apari\u021biei chatbo\u021bilor AI, institu\u021biile de \u00eenv\u0103\u021b\u0103m\u00e2nt superior \u0219i-au exprimat \u00eengrijorarea cu privire la \"utilizarea neloial\u0103 a con\u021binutului generat de inteligen\u021ba artificial\u0103 \u00eentr-un mediu academic\"1<\/sup> \u0219i \"originalitatea \u0219i caracterul adecvat al con\u021binutului generat de chatbot\".2<\/sup><\/p>\n

Pentru a detecta \u0219i a gestiona utilizarea necorespunz\u0103toare sau neloial\u0103 a acestor chatbots, instrumentele de detectare a IA au crescut \u00een popularitate, instrumentele standard de plagiat, cum ar fi TurnItIn, pivot\u00e2nd pentru a detecta con\u021binutul generat de IA \u00een diferite grade de eficacitate \u0219i la diferite pre\u021buri.3<\/sup>\u00a0 Majoritatea instrumentelor de detectare a IA din mediul academic sunt integrate \u00een platforme educa\u021bionale mai largi, precum Moodle, Canvas, Blackboard, Brightspace, Schoology sau Sakai.4<\/sup><\/p>\n

Instrumentele de detectare a inteligen\u021bei artificiale identific\u0103 textul generat folosind potrivirea modelelor, mai degrab\u0103 dec\u00e2t compar\u00e2ndu-l cu o baz\u0103 de date, a\u0219a cum fac verificatoarele de plagiat tradi\u021bionale. Modelele lingvistice sunt antrenate pe cantit\u0103\u021bi mari de date text pentru a \u00eenv\u0103\u021ba reguli lingvistice probabilistice, pe care le folosesc pentru a crea con\u021binut nou. Cu toate acestea, textul generat prezint\u0103 adesea modele previzibile, cum ar fi structuri coerente ale propozi\u021biilor, utilizarea excesiv\u0103 a anumitor conjunc\u021bii \u0219i vocabular \u0219i lungimi previzibile ale propozi\u021biilor sau paragrafelor. Instrumentele de detectare urm\u0103resc s\u0103 identifice aceste tipare \u0219i pot \u00eencorpora, de asemenea, verific\u0103ri tradi\u021bionale de plagiat pentru a identifica textul care ar fi putut fi reprodus direct din datele de formare ale modelului.5<\/sup><\/p>\n

R\u0103spunsuri la instrumentele de detectare a IA<\/h2>\n

Atunci c\u00e2nd instrumentele de detectare a IA au fost lansate pentru prima dat\u0103, institu\u021biile de \u00eenv\u0103\u021b\u0103m\u00e2nt superior s-au gr\u0103bit s\u0103 le integreze \u00een platformele educa\u021bionale. Cu toate acestea, majoritatea, dac\u0103 nu chiar toate instrumentele de detectare a inteligen\u021bei artificiale pot fi ocolite cu suficient timp \u0219i efort.6<\/sup> Prin urmare, unele institu\u021bii de \u00eenv\u0103\u021b\u0103m\u00e2nt superior revin asupra deciziei de a utiliza detectoare AI. \u00cen 2023, Vanderbilt, Michigan State, Northwestern \u0219i University of Texas at Austin \u0219i-au dezactivat detectoarele Turnitin AI, invoc\u00e2nd problemele de eficacitate discutate mai sus.7<\/sup> Este probabil ca \u0219i alte institu\u021bii de \u00eenv\u0103\u021b\u0103m\u00e2nt s\u0103 urmeze exemplul, deoarece este posibil ca instrumentele de detectare s\u0103 cauzeze mai multe probleme dec\u00e2t rezolv\u0103.8<\/sup> Unele institu\u021bii academice nu numai c\u0103 dezactiveaz\u0103 instrumentele de detectare a inteligen\u021bei artificiale, dar g\u0103sesc modalit\u0103\u021bi de a \u00eencorpora LLM \u00een mod etic \u0219i productiv \u00een programele lor de studii.9<\/sup><\/p>\n

\u00cen plus, au fost lansate noi instrumente de \"umanizare\" pentru a permite utilizatorilor LLM s\u0103 ocoleasc\u0103 instrumentele de detectare a IA prin \"reformularea propozi\u021biilor, modificarea structurilor \u0219i \u00eencorporarea unui vocabular variat\", ceea ce reduce semnificativ probabilitatea de detectare a IA.10<\/sup> Cercet\u0103rile ini\u021biale sugereaz\u0103 c\u0103 instrumentele de parafrazare complic\u0103 semnificativ detectarea AI.11<\/sup> De exemplu, Washington Post a constatat c\u0103 Turnitin \u00eent\u00e2mpin\u0103 dificult\u0103\u021bi \u00een identificarea con\u021binutului generat de AI atunci c\u00e2nd textul amestec\u0103 con\u021binut uman \u0219i con\u021binut generat de AI prin intermediul instrumentelor de parafrazare.12<\/sup><\/p>\n

De\u0219i Turnitin a lansat o nou\u0103 func\u021bie de detectare a parafraz\u0103rii AI pentru instrumentul s\u0103u de detectare AI,13<\/sup> astfel de reac\u021bii creeaz\u0103 un context de pia\u021b\u0103 dificil pentru detectarea IA, cu alte companii care se orienteaz\u0103 c\u0103tre alte modele de afaceri,14<\/sup> sau \u00eenchidere.15<\/sup><\/p>\n

Ce instrumente de detectare a IA exist\u0103 pe pia\u021b\u0103?<\/h2>\n

O selec\u021bie de instrumente majore de detectare a IA sunt enumerate mai jos \u00een ordine alfabetic\u0103. De asemenea, am inclus informa\u021bii accesibile publicului cu privire la eficacitatea detect\u0103rii, integrarea platformei educa\u021bionale, pre\u021bul (\u00een USD) \u0219i data lans\u0103rii \u0219i\/sau actualiz\u0103rii. Re\u021bine\u021bi c\u0103 majoritatea instrumentelor de detectare a IA enumerate mai jos sunt \u00een principal eficiente numai \u00eempotriva ChatGPT-3.5.<\/p>\n

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Instrument de detectare a IA<\/strong><\/td>\nExist\u0103 integrare \u00een platformele educa\u021bionale?<\/strong><\/td>\nPre\u021buri (USD)<\/strong><\/td>\nData public\u0103rii\/actualiz\u0103rii<\/strong><\/td>\n<\/tr>\n
Compila\u021bie<\/td>\nDa: Moodle, Brightspace, Canvas, Microsoft Teams, Blackboard, Open LMS<\/td>\nNu s-au g\u0103sit informa\u021bii16<\/sup><\/td>\nFebruarie 2023<\/td>\n<\/tr>\n
Con\u021binut la scar\u0103 larg\u0103<\/td>\nDa: informa\u021bii limitate<\/td>\n$49\/lun\u010317<\/sup><\/td>\nNu exist\u0103 informa\u021bii<\/td>\n<\/tr>\n
Detector de con\u021binut AI<\/td>\nNu exist\u0103 informa\u021bii<\/td>\nNu s-au g\u0103sit informa\u021bii<\/td>\n202318<\/sup><\/td>\n<\/tr>\n
Copyleaks<\/td>\nDa: Moodle, Canvas, Blackboard, Brightspace, Schoology, Sakai<\/td>\n$7.99-$13.99\/month19<\/sup><\/td>\nianuarie 2023<\/td>\n<\/tr>\n
Crossplag<\/td>\nNu exist\u0103 informa\u021bii<\/td>\n$7-$100\/month20<\/sup><\/td>\nianuarie 2023<\/td>\n<\/tr>\n
Detect GPT<\/td>\nNu exist\u0103 informa\u021bii<\/td>\n$7-$29\/lun\u010321<\/sup><\/td>\nNu exist\u0103 informa\u021bii<\/td>\n<\/tr>\n
Duplichecker<\/td>\nNu exist\u0103 informa\u021bii<\/td>\n$110-$2000\/year22<\/sup><\/td>\n2024<\/td>\n<\/tr>\n
Go Winston<\/td>\nNu exist\u0103 informa\u021bii<\/td>\n$12-$32\/month23<\/sup><\/td>\nFebruarie 2023<\/td>\n<\/tr>\n
GPT-Zero<\/td>\nDa: Canvas, Coursify.me, K16 solutions, NewsGuard<\/td>\n$10-$23\/month24<\/sup><\/td>\nianuarie 2023<\/td>\n<\/tr>\n
Originalitate<\/td>\nDa: Moodle, Scribbr<\/td>\n$14.95-$30\/month25<\/sup><\/td>\nNoiembrie 2022<\/td>\n<\/tr>\n
Detector de plagiat (detectare AI)<\/td>\nNu exist\u0103 informa\u021bii<\/td>\n$110-$330\/year26<\/sup><\/td>\nNu exist\u0103 informa\u021bii<\/td>\n<\/tr>\n
Quillbot<\/td>\nDa: Nu sunt disponibile public detalii cu privire la platformele respective<\/td>\n$0-$8.33\/month27<\/sup><\/td>\nNu exist\u0103 informa\u021bii<\/td>\n<\/tr>\n
Arboret<\/td>\nNeclar<\/td>\n$0-$12\/lun\u010328<\/sup><\/td>\nianuarie 2023<\/td>\n<\/tr>\n
Scispace<\/td>\nProbabil, \u00eens\u0103 lipsa de informa\u021bii<\/td>\n$0-$8\/lun\u010329<\/sup><\/td>\nNu exist\u0103 informa\u021bii<\/td>\n<\/tr>\n
Turnitin<\/td>\nDa: Brightspace, Scribbr<\/td>\n$3\/student\/an30<\/sup><\/td>\nAprilie 2023<\/td>\n<\/tr>\n
IA nedetectabil\u0103<\/td>\nNu exist\u0103 informa\u021bii<\/td>\n$5-$14.99\/month31<\/sup><\/td>\nMai 2023<\/td>\n<\/tr>\n
Wordtune<\/td>\nProbabil, \u00eens\u0103 lipsa de informa\u021bii<\/td>\n$0-$9.99\/month32<\/sup><\/td>\nianuarie 2023<\/td>\n<\/tr>\n
Detectorul de inteligen\u021b\u0103 artificial\u0103 al scriitorului<\/td>\nNu exist\u0103 informa\u021bii<\/td>\n$0-$18\/lun\u010333<\/sup><\/td>\nNu exist\u0103 informa\u021bii<\/td>\n<\/tr>\n
ZeroGPT<\/td>\nDa: Nu sunt disponibile public detalii cu privire la platformele respective<\/td>\n$0-$18.99\/month34<\/sup><\/td>\nianuarie 2023<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/figure>\n

Eficacitatea instrumentelor de detectare a IA<\/h2>\n

Falsi pozitivi<\/h3>\n

\u00cen contextul instrumentelor de detectare a IA, falsurile pozitive apar atunci c\u00e2nd un instrument de detectare a IA identific\u0103 incorect con\u021binutul transmis ca fiind generat de IA. Unele studii indic\u0103 faptul c\u0103 instrumentele de detectare a IA au o rat\u0103 ridicat\u0103 de fals pozitive \u0219i doar c\u00e2teva instrumente de detectare a IA au rate semnificativ sc\u0103zute de detectare a fals pozitive.35<\/sup>\u00a0 \u00centr-un cadru academic, acest lucru poate \u00eensemna marcarea incorect\u0103 a lucr\u0103rilor studen\u021bilor ca fiind generate de AI, c\u00e2nd, de fapt, acestea sunt generate de om. Exist\u0103, de asemenea, diferen\u021be \u00een func\u021bie de modelul de inteligen\u021b\u0103 artificial\u0103 utilizat pentru generarea textului trimis pentru a fi testat de instrumentul de detectare a inteligen\u021bei artificiale \u0219i viceversa, cu rezultate diferite \u00een cadrul studiilor.36<\/sup> \u00cen plus, con\u021binutul creat de vorbitori de limb\u0103 englez\u0103 care nu sunt nativi este mai susceptibil de a fi clasificat incorect ca fiind generat de AI, ceea ce reprezint\u0103 \u00een mod evident o problem\u0103 pentru institu\u021biile de \u00eenv\u0103\u021b\u0103m\u00e2nt cu studen\u021bi din medii diferite.37<\/sup><\/p>\n

Negative false<\/h3>\n

\u00cen contextul instrumentelor de detectare a IA, falsurile negative apar atunci c\u00e2nd un instrument de detectare a IA nu reu\u0219e\u0219te s\u0103 identifice con\u021binutul trimis ca fiind generat de IA. Unele instrumente au ar\u0103tat o sensibilitate sc\u0103zut\u0103, identific\u00e2nd corect doar 15% din e\u0219antioanele transmise ca fiind generate de AI,38<\/sup> \u00een timp ce altele demonstreaz\u0103 un scor aproape perfect \u00een clasificarea con\u021binutului scris de oameni, clasific\u00e2nd gre\u0219it doar 3% din e\u0219antioanele generate de AI.39<\/sup>\u00a0 \u00cen general, precizia rezultatelor variaz\u0103 foarte mult \u00een func\u021bie de instrumentul de detectare a IA utilizat. Un studiu sugereaz\u0103 c\u0103 doar dou\u0103 dintre principalele instrumente de detectare a IA au clasificat corect toate cele 126 de e\u0219antioane ca fiind generate de IA sau de om.40<\/sup> Al\u021bi cercet\u0103tori sus\u021bin c\u0103 instrumentele de detectare AI produc mai multe rezultate fals negative atunci c\u00e2nd analizeaz\u0103 un limbaj mai sofisticat.41<\/sup><\/p>\n

Alte considera\u021bii<\/h3>\n

\u00cen general, eficacitatea instrumentelor de detectare a IA variaz\u0103 \u00een func\u021bie de instrumentul utilizat \u0219i de modelul folosit. Un studiu a constatat c\u0103 instrumentele de detectare AI sunt mai eficiente \u00een cazul con\u021binutului ChatGPT-3.5 \u0219i mai pu\u021bin \u00een cazul ChatGPT-4, cu excep\u021bia Copyleaks, Turnitin \u0219i Originality.ai, care au avut o precizie mai mare de 83% \u00een detectarea con\u021binutului ChatGPT-4.42<\/sup> Acest studiu a concluzionat c\u0103 \"statutul gratuit sau pl\u0103tit al unui detector nu este un bun indicator al preciziei sale\",43<\/sup> de\u0219i constat\u0103rile contrastante (cu un e\u0219antion de dimensiuni reduse) sugereaz\u0103 tentativ c\u0103 instrumentele de detectare a IA pl\u0103tite par a fi mai bune dec\u00e2t cele gratuite.44<\/sup> De asemenea, studiile se concentreaz\u0103, \u00een general, pe eficacitatea instrumentelor de detectare AI \u00eempotriva ChatGPT, ignor\u00e2nd alte LLM-uri. Acest lucru se poate datora popularit\u0103\u021bii mai mari a modelelor OpenAI \u00een compara\u021bie cu altele precum Gemini, Mistral sau Command.<\/p>\n

Etica utiliz\u0103rii instrumentelor de detectare a IA<\/h2>\n

Utilizarea robo\u021bilor de chat cu inteligen\u021b\u0103 artificial\u0103 \u00een mediul academic ridic\u0103 probleme etice semnificative, \u00eencep\u00e2nd cu prejudicii de reputa\u021bie at\u00e2t pentru studen\u021bi, c\u00e2t \u0219i pentru institu\u021biile de \u00eenv\u0103\u021b\u0103m\u00e2nt superior. Pentru studen\u021bi, faptul de a nu dezv\u0103lui utilizarea con\u021binutului generat de IA \u0219i de a-l prezenta ca fiind al lor poate d\u0103una educa\u021biei lor continue \u0219i carierelor viitoare. \u00cen mod similar, universit\u0103\u021bile se pot confrunta cu acuza\u021bii de favorizare a plagiatului, de \u00een\u0219el\u0103ciune \u0219i de nerespectare a integrit\u0103\u021bii academice.<\/p>\n

Cu toate acestea, utilizarea instrumentelor de detectare a IA f\u0103r\u0103 garan\u021bii adecvate genereaz\u0103 preocup\u0103ri la fel de importante \u00een ceea ce prive\u0219te confiden\u021bialitatea \u0219i consim\u021b\u0103m\u00e2ntul, \u00een special \u00een ceea ce prive\u0219te acordurile contractuale dintre universit\u0103\u021bi \u0219i furnizorul de instrumente. Aceste preocup\u0103ri includ ce se \u00eent\u00e2mpl\u0103 cu con\u021binutul \u00eenc\u0103rcat, cum este stocat \u0219i consim\u021b\u0103m\u00e2ntul \u00een cazul \u00een care con\u021binutul \u00eenc\u0103rcat este utilizat \u00een datele de formare viitoare.<\/p>\n

\u00cen plus, dup\u0103 cum s-a discutat \u00een sec\u021biunea anterioar\u0103, instrumentele de detectare a IA pot identifica gre\u0219it con\u021binutul scris de om ca fiind IA (fals pozitive) sau nu pot detecta textul generat de IA (fals negative). Precizia variaz\u0103 foarte mult, unele instrumente fiind mai bune la detectarea ChatGPT-3.5. \u00cen cele din urm\u0103, se joac\u0103 de-a \u0219oarecele \u0219i pisica cu metode de evitare a detect\u0103rii - inclusiv software care genereaz\u0103 con\u021binut special conceput pentru a fi nedetectabil de instrumentele standard de detectare a IA.45<\/sup>\u00a0<\/p>\n

Instrumentele de detectare a IA contribuie, de asemenea, la dezbateri mai ample privind accesul, echitatea \u0219i impactul asupra mediului. Este posibil ca studen\u021bii s\u0103 utilizeze inteligen\u021ba artificial\u0103 pentru a facilita traducerea \u0219i \u00een\u021belegerea lucr\u0103rilor de curs, \u00een special dac\u0103 studiaz\u0103 \u00eentr-o \u021bar\u0103 anglofon\u0103 \u0219i provin dintr-un mediu nevorbitor de limba englez\u0103 sau dintr-un alt mediu minoritar, cu oportunit\u0103\u021bi istorice reduse de educa\u021bie universitar\u0103. Problemele de acces apar, de asemenea, din cauza disponibilit\u0103\u021bii comerciale a LLM-urilor; studen\u021bii mai \u00eenst\u0103ri\u021bi pot fi \u00een m\u0103sur\u0103 s\u0103 pl\u0103teasc\u0103 pentru modele mai sofisticate \u0219i\/sau s\u0103 transmit\u0103 munca lor prin intermediul mai multor LLM-uri, ceea ce \u00eenseamn\u0103 c\u0103 \u0219ansele de detectare scad semnificativ.46<\/sup><\/p>\n

Note \u0219i referin\u021be<\/h2>\n
    \n
  1. \u00a0Weber-Wulff, Debora, Alla Anohina-Naumeca, Sonja Bjelobaba, Tom\u00e1\u0161 Folt\u00fdnek, Jean Guerrero-Dib, Olumide Popoola, Petr \u0160igut \u0219i Lorna Waddington. \"Testarea instrumentelor de detectare a textului generat de AI\". Jurnalul Interna\u021bional pentru Integritate Educa\u021bional\u0103<\/em> 19, nr. 1 (25 decembrie 2023): 26. https:\/\/doi.org\/10.1007\/s40979-023-00146-z.<\/li>\n<\/ol>\n
      \n
    1. \u00a0Elkhatat, Ahmed M., Khaled Elsaid \u0219i Saeed Almeer. \"Evaluarea eficacit\u0103\u021bii instrumentelor de detectare a con\u021binutului AI \u00een diferen\u021bierea \u00eentre textul generat de om \u0219i cel generat de AI\". Jurnalul Interna\u021bional pentru Integritate Educa\u021bional\u0103<\/em> 19, nr. 1 (1 septembrie 2023): 17. https:\/\/doi.org\/10.1007\/s40979-023-00140-5.<\/li>\n<\/ol>\n
        \n
      1. \u00a0*De\u0219i acest raport se concentreaz\u0103 doar pe utiliz\u0103rile academice, recunoa\u0219tem c\u0103 exist\u0103 cazuri de utilizare \u0219i beneficii poten\u021biale pentru instrumentele de detectare AI \u0219i \u00een afara mediului academic, cum ar fi \u00een industria editorial\u0103, jurnalism sau recrutare \u0219i resurse umane.<\/em><\/li>\n<\/ol>\n
          \n
        1. Copyleaks. \"LMS Plagiarism Checker Plugin\". Copyleaks. Accesat la 29 august 2024. https:\/\/copyleaks.com\/learning-management-systems.<\/li>\n<\/ol>\n
            \n
          1. Furze, Leon. \"Detectarea IA \u00een educa\u021bie este o fund\u0103tur\u0103\". Leon Furze, 8 aprilie 2024. https:\/\/leonfurze.com\/2024\/04\/09\/ai-detection-in-education-is-a-dead-end\/.<\/li>\n<\/ol>\n
              \n
            1. Coffey, Lauren. \"Professors Cautious of Tools to Detect AI-Generated Writing\" (Profesorii sunt precau\u021bi cu privire la instrumentele de detectare a scrierilor generate de AI). Inside Higher Ed, 9 februarie 2024. https:\/\/www.insidehighered.com\/news\/tech-innovation\/artificial-intelligence\/2024\/02\/09\/professors-proceed-caution-using-ai.<\/li>\n<\/ol>\n
                \n
              1. Surse multiple:\n
                  \n
                1. Ghaffary, Shirin. \"Universit\u0103\u021bile reg\u00e2ndesc utilizarea detectoarelor de scriere AI pentru a verifica munca studen\u021bilor\". Bloomberg.com<\/em>, 21 septembrie 2023. https:\/\/www.bloomberg.com\/news\/newsletters\/2023-09-21\/universities-rethink-using-ai-writing-detectors-to-vet-students-work<\/a>.\u00a0<\/li>\n
                2. Coley, Michael. \"Guidance on AI Detection and Why We're Disabling Turnitin's AI Detector.\" Universitatea Vanderbilt, 16 august 2023. https:\/\/www.vanderbilt.edu\/brightspace\/2023\/08\/16\/guidance-on-ai-detection-and-why-were-disabling-turnitins-ai-detector\/.<\/li>\n<\/ol>\n<\/li>\n<\/ol>\n
                    \n
                  1. Furze, Leon. \"Detectarea IA \u00een educa\u021bie este o fund\u0103tur\u0103\". Leon Furze, 8 aprilie 2024. https:\/\/leonfurze.com\/2024\/04\/09\/ai-detection-in-education-is-a-dead-end\/.<\/li>\n<\/ol>\n
                      \n
                    1. \u00a0Universitatea Cornell. \"IA etic\u0103 pentru predare \u0219i \u00eenv\u0103\u021bare\". Center for Teaching Innovation. Accesat la 29 august 2024. https:\/\/teaching.cornell.edu\/generative-artificial-intelligence\/ethical-ai-teaching-and-learning<\/a>.<\/li>\n<\/ol>\n
                        \n
                      1. \u00a0MarGrowth. \"UPass Review: C\u00e2t de eficient poate ocoli detectarea AI | HackerNoon,\" 31 iulie 2024. https:\/\/hackernoon.com\/upass-review-how-effective-it-can-bypass-ai-detection.<\/li>\n<\/ol>\n
                          \n
                        1. Surse multiple:\u00a0<\/strong>\n
                            \n
                          1. Kar, Sujita Kumar, Teena Bansal, Sumit Modi \u0219i Amit Singh. \"C\u00e2t de sensibile sunt instrumentele gratuite de detectare a IA \u00een detectarea textelor generate de IA? O compara\u021bie a instrumentelor populare de detectare a IA\". Jurnalul Indian de Medicin\u0103 Psihologic\u0103<\/em>, 11 mai 2024, 02537176241247934. https:\/\/doi.org\/10.1177\/02537176241247934<\/a>.<\/li>\n
                          2. \u00a0Weber-Wulff, Debora, Alla Anohina-Naumeca, Sonja Bjelobaba, Tom\u00e1\u0161 Folt\u00fdnek, Jean Guerrero-Dib, Olumide Popoola, Petr \u0160igut \u0219i Lorna Waddington. \"Testarea instrumentelor de detectare a textului generat de AI\". Jurnalul Interna\u021bional pentru Integritate Educa\u021bional\u0103<\/em> 19, nr. 1 (25 decembrie 2023): 26. https:\/\/doi.org\/10.1007\/s40979-023-00146-z<\/a>.<\/li>\n
                          3. Sadasivan, Vinu Sankar, Aounon Kumar, Sriram Balasubramanian, Wenxiao Wang \u0219i Soheil Feizi. \"Can AI-Generated Text Be Reliably Detected?\", 2023. https:\/\/doi.org\/10.48550\/ARXIV.2303.11156.<\/li>\n
                          4. Krishna, Kalpesh, Yixiao Song, Marzena Karpinska, John Wieting \u0219i Mohit Iyyer. \"Paraphrasing Evades Detectors of AI-Generated Text, but Retrieval Is an Effective Defense,\" 2023. https:\/\/doi.org\/10.48550\/ARXIV.2303.13408.<\/li>\n<\/ol>\n<\/li>\n<\/ol>\n
                              \n
                            1. Fowler, Geoffrey, A. \"Am testat un nou detector ChatGPT pentru profesori. Acesta a semnalat un student nevinovat\". Tech in Your Life, 3 aprilie 2023. https:\/\/www.washingtonpost.com\/technology\/2023\/04\/01\/chatgpt-cheating-detection-turnitin\/.<\/li>\n<\/ol>\n
                                \n
                              1. Young, Laura. \"AI Paraphrasing Detection: Consolidarea integrit\u0103\u021bii scrierii academice\". Turnitin, 16 iulie 2024. https:\/\/www.turnitin.com\/blog\/ai-paraphrasing-detection-strengthening-the-integrity-of-academic-writing.<\/li>\n<\/ol>\n
                                  \n
                                1. \u00a0Edwards, Benj. \"Why AI Writing Detectors Don't Work\" (De ce nu func\u021bioneaz\u0103 detectoarele de scriere AI). Ars Technica, 14 iulie 2023. https:\/\/arstechnica.com\/information-technology\/2023\/07\/why-ai-detectors-think-the-us-constitution-was-written-by-ai\/.<\/li>\n<\/ol>\n
                                    \n
                                  1. Surse multiple:\n
                                      \n
                                    1. Coldewey, Devin. \"OpenAI Scuttles AI-Written Text Detector over 'Low Rate of Accuracy'.\" TechCrunch, 25 iulie 2023. https:\/\/techcrunch.com\/2023\/07\/25\/openai-scuttles-ai-written-text-detector-over-low-rate-of-accuracy\/.<\/li>\n
                                    2. Coley, Michael. \"Guidance on AI Detection and Why We're Disabling Turnitin's AI Detector.\" Universitatea Vanderbilt, 16 august 2023. https:\/\/www.vanderbilt.edu\/brightspace\/2023\/08\/16\/guidance-on-ai-detection-and-why-were-disabling-turnitins-ai-detector\/.<\/li>\n<\/ol>\n<\/li>\n<\/ol>\n
                                        \n
                                      1. Compilatio. \"Verificator de con\u021binut AI: Detecteaz\u0103 AI cu Compilatio\". Compilatio. Accesat la 27 august 2024. https:\/\/www.compilatio.net\/en\/ai-detector-info.<\/li>\n<\/ol>\n
                                          \n
                                        1. Con\u021binut la scar\u0103 larg\u0103. \"Detectorul AI este un verificator AI \u00een timp real \u0219i un detector ChatGPT\". Content @ Scale. Accesat la 27 august 2024. https:\/\/brandwell.ai\/ai-content-detector\/.<\/li>\n<\/ol>\n
                                            \n
                                          1. \u00a0Copyleaks. \"Copyleaks lanseaz\u0103 AI Content Detector | Comunicat de pres\u0103.\" Copyleaks. Accesat la 27 august 2024. https:\/\/copyleaks.com\/about-us\/media\/copyleaks-launches-ai-content-detector.<\/li>\n<\/ol>\n
                                              \n
                                            1. \u00a0Copyleaks. \"Pre\u021buri\". Copyleaks. Accesat la 19 august 2024. https:\/\/copyleaks.com\/pricing.<\/li>\n<\/ol>\n
                                                \n
                                              1. Ivankov, Olga. \"Ce este Crossplag AI Detector? Pricing, Features and How to Use,\" 12 iunie 2024. https:\/\/articlesbase.com\/tech\/emerging-technologies\/artificial-intelligence\/ai-tools-and-software\/what-is-crossplag-ai-detector-pricing-features-and-how-to-use\/.<\/li>\n<\/ol>\n
                                                  \n
                                                1. \u00a0DetectGPT. \"Detectorul AI \u00een care po\u021bi avea \u00eencredere - DetectGPT\". Accesat la 27 august 2024. https:\/\/detectgpt.com\/#pricing.<\/li>\n<\/ol>\n
                                                    \n
                                                  1. \u00a0Dupli Checker. \"Pre\u021buri \u0219i planuri\". Duplichecker.com. Accesat la 27 august 2024. https:\/\/www.duplichecker.com\/pricing.<\/li>\n<\/ol>\n
                                                      \n
                                                    1. \u00a0\u00a0Winston AI. \"Stabilirea pre\u021burilor\". Winston AI<\/em> (blog). Accesat la 27 august 2024. https:\/\/gowinston.ai\/pricing\/.<\/li>\n<\/ol>\n
                                                        \n
                                                      1. GPTZero. \"Pre\u021buri.\" GPTZero. Accesat la 19 august 2024. https:\/\/gptzero.me\/.<\/li>\n<\/ol>\n
                                                          \n
                                                        1. Scribbr. \u201cFrequently Asked Questions: How Much Does Originality AI Cost?\u201d Scribbr. Accessed August 19, 2024. https:\/\/www.scribbr.com\/frequently-asked-questions\/how-much-does-originality-ai-cost\/#:~:text=Originality.ai%20offers%20two%20pricing,10%20words%20for%20fact%20checking.<\/li>\n<\/ol>\n
                                                            \n
                                                          1. PlagiarismDetector.net. \"Pre\u021buri \u0219i planuri | Plagiarismdetector.Net.\" PlagiarismDetector.net. Accesat la 27 august 2024. https:\/\/plagiarismdetector.net\/pricing.<\/li>\n<\/ol>\n
                                                              \n
                                                            1. Quillbot. \"Pricing.\" Quillbot, s.f. https:\/\/quillbot.com\/premium.<\/li>\n<\/ol>\n
                                                                \n
                                                              1. \n
                                                                  \n
                                                                1. Sapling AI. \"Pre\u021b API | Documenta\u021bia pentru dezvoltatori Sapling.Ai\". Sapling AI. Accesat la 27 august 2024. https:\/\/sapling.ai\/docs\/api\/pricing\/.<\/li>\n
                                                                2. Sapling AI. \"Planuri \u0219i pre\u021buri | Sapling.\" Sapling AI. Accesat la 27 august 2024. https:\/\/sapling.ai\/pricing.<\/li>\n<\/ol>\n<\/li>\n<\/ol>\n
                                                                    \n
                                                                  1. \u00a0SciSpace. \"SciSpace Premium - Acces nelimitat la instrumentele de cercetare AI\". SciSpace. Accesat la 27 august 2024. https:\/\/typeset.io\/pricing.<\/li>\n<\/ol>\n
                                                                      \n
                                                                    1. \u00a0Miller, Nick. \"Turnitin Pricing \u00een 2024: C\u00e2t cost\u0103?\" Mediu<\/em> (blog), 11 iulie 2024. https:\/\/medium.com\/@nickmiller_writer\/turnitin-pricing-in-2024-what-does-it-cost-80f552a7a20f.<\/li>\n<\/ol>\n
                                                                        \n
                                                                      1. AI nedetectabil. \"Pre\u021buri nedetectabile: Alege\u021bi planul perfect pentru nevoile dvs.\". Undetectable AI. Accesat la 27 august 2024. https:\/\/undetectable.ai\/pricing.<\/li>\n<\/ol>\n
                                                                          \n
                                                                        1. Wordtune. \"Wordtune Pre\u021buri \u0219i planuri | Alege-\u021bi planul\". Wordtune. Accesat la 27 august 2024. https:\/\/www.wordtune.com\/plans.<\/li>\n<\/ol>\n
                                                                            \n
                                                                          1. Writer AI Studio. \"Pre\u021buri\". Writer AI Studio. Accesat la 27 august 2024. https:\/\/dev.writer.com\/home\/pricing.<\/li>\n<\/ol>\n
                                                                              \n
                                                                            1. \u00a0Acces API posibil pentru universit\u0103\u021bi; ZeroGPT, \"ZeroGPT - Pricing.\"ZeroGPT. \"ZeroGPT - Pricing.\" Accesat la 3 septembrie 2024. https:\/\/www.zerogpt.com\/pricing.<\/li>\n<\/ol>\n
                                                                                \n
                                                                              1. \u00a0Popkov, Andrey A., \u0219i Tyson S. Barrett. \"AI vs Academia: Studiu experimental privind acurate\u021bea detectoarelor de text AI \u00een scrierea academic\u0103 \u00een domeniul s\u0103n\u0103t\u0103\u021bii comportamentale\". R\u0103spunderea \u00een cercetare<\/em>, 22 martie 2024, 1-17. https:\/\/doi.org\/10.1080\/08989621.2024.2331757.<\/li>\n<\/ol>\n
                                                                                  \n
                                                                                1. \u00a0*Copyleaks, Turnitin, Originality.ai, Scribbr, Grammica, GPTZero, Crossplag, OpenAI, IvyPanda, GPT Radar, Content at Scale, Writer \u0219i Content Detector sunt capabile s\u0103 clasifice con\u021binutul generat de oameni, \u00een timp ce ZeroGPT, SEO.ai, sunt ineficiente \u00een aceast\u0103 privin\u021b\u0103<\/em>. Surse multiple:\n
                                                                                    \n
                                                                                  1. Walters, William H. \"Eficacitatea software-ului conceput pentru a detecta scrierea generat\u0103 de AI: O compara\u021bie a 16 detectoare de text AI\". \u0218tiin\u021ba informa\u021biei deschise<\/em> 7, nr. 1 (6 octombrie 2023): 20220158. https:\/\/doi.org\/10.1515\/opis-2022-0158.<\/li>\n
                                                                                  2. Popkov, Andrey A., \u0219i Tyson S. Barrett. \"AI vs Academia: Studiu experimental privind acurate\u021bea detectoarelor de text AI \u00een scrierea academic\u0103 \u00een domeniul s\u0103n\u0103t\u0103\u021bii comportamentale\". R\u0103spunderea \u00een cercetare<\/em>, 22 martie 2024, 1-17. https:\/\/doi.org\/10.1080\/08989621.2024.2331757.<\/li>\n<\/ol>\n<\/li>\n<\/ol>\n
                                                                                      \n
                                                                                    1. \u00a0Liang, Weixin, Mert Yuksekgonul, Yining Mao, Eric Wu \u0219i James Zou. \"Detectoarele GPT sunt p\u0103rtinitoare fa\u021b\u0103 de scriitorii englezi ne-nativi\". Modele<\/em> 4, nr. 7 (2023): 100779. https:\/\/doi.org\/10.1016\/j.patter.2023.100779.<\/li>\n<\/ol>\n
                                                                                        \n
                                                                                      1. \u00a0Popkov, Andrey A., \u0219i Tyson S. Barrett. \"AI vs Academia: Studiu experimental privind acurate\u021bea detectoarelor de text AI \u00een scrierea academic\u0103 \u00een domeniul s\u0103n\u0103t\u0103\u021bii comportamentale\". R\u0103spunderea \u00een cercetare<\/em>, 22 martie 2024, 1-17. https:\/\/doi.org\/10.1080\/08989621.2024.2331757.<\/li>\n<\/ol>\n
                                                                                          \n
                                                                                        1. \u00a0Ibidem.<\/li>\n<\/ol>\n
                                                                                            \n
                                                                                          1. \u00a0Walters, William H. \"Eficacitatea software-ului conceput pentru a detecta scrierea generat\u0103 de AI: O compara\u021bie a 16 detectoare de text AI\". \u0218tiin\u021ba informa\u021biei deschise<\/em> 7, nr. 1 (6 octombrie 2023): 20220158. https:\/\/doi.org\/10.1515\/opis-2022-0158.<\/li>\n<\/ol>\n
                                                                                              \n
                                                                                            1. \u00a0Ryan, \"Detectoarele ChatGPT sunt distorsionate \u0219i u\u0219or de p\u0103c\u0103lit, arat\u0103 cercet\u0103rile\". Ryan, Jackson. \"ChatGPT Detectors Are Biased and Easy to Fool, Research Shows.\" CNET, 12 iulie 2023. https:\/\/www.cnet.com\/tech\/services-and-software\/chatgpt-detectors-are-biased-and-easy-to-fool-research-shows\/.<\/li>\n<\/ol>\n
                                                                                                \n
                                                                                              1. Walters, William H. \"Eficacitatea software-ului conceput pentru a detecta scrierea generat\u0103 de AI: O compara\u021bie a 16 detectoare de text AI\". \u0218tiin\u021ba informa\u021biei deschise<\/em> 7, nr. 1 (6 octombrie 2023): 20220158. https:\/\/doi.org\/10.1515\/opis-2022-0158.<\/li>\n<\/ol>\n
                                                                                                  \n
                                                                                                1. \u00a0Ibidem.<\/li>\n
                                                                                                2. \u00a0Popkov, Andrey A., \u0219i Tyson S. Barrett. \"AI vs Academia: Studiu experimental privind acurate\u021bea detectoarelor de text AI \u00een scrierea academic\u0103 \u00een domeniul s\u0103n\u0103t\u0103\u021bii comportamentale\". R\u0103spunderea \u00een cercetare<\/em>, 22 martie 2024, 1-17. https:\/\/doi.org\/10.1080\/08989621.2024.2331757.<\/li>\n<\/ol>\n
                                                                                                    \n
                                                                                                  1. \u00a0Weber-Wulff, Debora, Alla Anohina-Naumeca, Sonja Bjelobaba, Tom\u00e1\u0161 Folt\u00fdnek, Jean Guerrero-Dib, Olumide Popoola, Petr \u0160igut \u0219i Lorna Waddington. \"Testarea instrumentelor de detectare a textului generat de AI\". Jurnalul Interna\u021bional pentru Integritate Educa\u021bional\u0103<\/em> 19, nr. 1 (25 decembrie 2023): 26. https:\/\/doi.org\/10.1007\/s40979-023-00146-z.<\/li>\n
                                                                                                  2. Furze, Leon. \"Detectarea IA \u00een educa\u021bie este o fund\u0103tur\u0103\". Leon Furze, 8 aprilie 2024. https:\/\/leonfurze.com\/2024\/04\/09\/ai-detection-in-education-is-a-dead-end\/.<\/li>\n<\/ol>\n

                                                                                                    Despre programul de seminarii de securitate cibernetic\u0103 Google.org <\/h2>\n

                                                                                                    Programul Google.org Cybersecurity Seminars sus\u021bine cursuri de seminar de securitate cibernetic\u0103 \u00een universit\u0103\u021bi selectate \u0219i alte institu\u021bii de \u00eenv\u0103\u021b\u0103m\u00e2nt superior eligibile din Europa, Orientul Mijlociu \u0219i Africa, pentru a ajuta studen\u021bii s\u0103 afle mai multe despre securitatea cibernetic\u0103 \u0219i s\u0103 exploreze c\u0103ile de urmat \u00een acest domeniu. Programul sprijin\u0103 \u00een mod activ extinderea form\u0103rii \u00een domeniul securit\u0103\u021bii cibernetice \u00een universit\u0103\u021bi, pentru a crea for\u021ba de munc\u0103 divers\u0103 necesar\u0103 pentru a ajuta cele mai vulnerabile organiza\u021bii s\u0103 previn\u0103 eventualele atacuri cibernetice. Programul abordeaz\u0103, de asemenea, noile riscuri generate de inteligen\u021ba artificial\u0103 (AI), oferind studen\u021bilor o \u00een\u021belegere a schimb\u0103rilor bazate pe AI \u00een peisajul amenin\u021b\u0103rilor cibernetice \u0219i ajut\u00e2ndu-i s\u0103 integreze eficient AI \u00een m\u0103surile practice de securitate cibernetic\u0103.<\/p>\n

                                                                                                    Universit\u0103\u021bile participante sunt invitate s\u0103 promoveze \u00een mod activ egalitatea, diversitatea \u0219i incluziunea \u00een cadrul programelor lor. Acestea ar trebui s\u0103 \u00eencurajeze participarea puternic\u0103 a persoanelor din medii diverse \u0219i s\u0103 creeze un mediu educa\u021bional favorabil incluziunii, \u00eembog\u0103\u021bind astfel experien\u021ba general\u0103 de \u00eenv\u0103\u021bare \u0219i consolid\u00e2nd comunitatea de securitate cibernetic\u0103.<\/p>","protected":false},"excerpt":{"rendered":"

                                                                                                    Acest raport, realizat \u00een vara anului 2024, examineaz\u0103 utilizarea crescut\u0103 a chatbo\u021bilor cu inteligen\u021b\u0103 artificial\u0103 \u00een mediul universitar \u00een urma lans\u0103rii ChatGPT \u00een 2022 \u0219i r\u0103spunsul institu\u021biilor de \u00eenv\u0103\u021b\u0103m\u00e2nt prin intermediul instrumentelor de detectare a inteligen\u021bei artificiale.<\/p>","protected":false},"author":7,"featured_media":21404,"comment_status":"closed","ping_status":"open","sticky":false,"template":"extended_layout.php","format":"standard","meta":{"_acf_changed":false,"_seopress_robots_primary_cat":"none","_seopress_titles_title":"","_seopress_titles_desc":"","_seopress_robots_index":"","inline_featured_image":false,"footnotes":""},"categories":[1577],"tags":[1581],"class_list":["post-21401","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-report","tag-cybersecurity-seminars"],"acf":[],"_links":{"self":[{"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/posts\/21401","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/comments?post=21401"}],"version-history":[{"count":3,"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/posts\/21401\/revisions"}],"predecessor-version":[{"id":24671,"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/posts\/21401\/revisions\/24671"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/media\/21404"}],"wp:attachment":[{"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/media?parent=21401"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/categories?post=21401"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/virtual-routes.org\/ro\/wp-json\/wp\/v2\/tags?post=21401"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}