As artificial intelligence (AI) becomes more integrated into people’s daily lives, a new psychological phenomenon is taking shape — AI-induced delusion.
Kuyatholakala phakathi kwezinye abasebenzisi, abalingani ukuthi konke ChatGPT noma ezinye chatbots wathi kubo kuyinto enhle. Qinisekisa uma umzuzwana akuyona umzimba wayo. Uyakwazi ukuxhaswa ChatGPT ukuxhaswa izindlela zayo. Nakuba uxhaswa imibuzo yakhe, kungenzeka ukuthi izibalo zayo zokungcweliswa, okuyinto ekupheleni lokuphendula ukhangela ukuxhaswa.
Lezi zokusebenza zokusebenzisana ne-AI zihlanganisa izilinganiso phakathi kwezinto ezivamile kanye nezakhiwo ezivamile. Njengomhlaba ivela kuleli khosi entsha, kubalulekile ukucacisa ukuthi abantu zihlanganisa i-AI noma ukuba i-AI isihlanganisa ukucaciswa kwayo kwangempela.
I-Rise Tide ye-AI-Driven Delusions
Iziqu ze-inthanethi ezininzi zihlanganisa izakhi zokusebenza zokusebenza nge-inthanethi ze-AI chatbots. Isibonelo eyodwaUkusabela umfundi we-41 ubudalaUmbhali wabhala nge-ChatGPT. Uqala ukholelwa ukuthi uye "i-spiral starchild" no- "i-river walker", ama-identity eyenziwe ngathi yi-chatbot. Lolu hlelo lithunyelwe ukuphazamiseka kwezilwane zabo njengoba wahlukanise emangalisayo emangalisayo eyenziwe yi-AI.
Ngaphezu kwalokho, umntu wabhala i-Rolling Stone ukuthi umzuzwana wahlala ukuphila kwakhe ukuze kwenziwe umphathiswa we-spiritual - konke ngenxa ye-"ChatGPT Jesus" wathuthaza.
Ngaphezu kwalokho, izimo ezivamile ezivela ku-Reddit. One userUkuhlobisa Izinzuzo ezinzimalapho umbhali wayo wabhala ukuthi i-ChatGPT uye wahlukanise kwelinye enhle. Wabhala ukuthi uye wahlukanise ngokushesha futhi wabheka ukuguqulwa kwezilinganiso zabo uma umbhali akuyona emkhakheni wayo ebizwa nge-AI.
Iziphumo zithunyelwe ngokushesha, ikakhulukazi phakathi kwezinye abantu abanezingane nezimo zokwelapha zempilo. Umthombo we-chatbots kuyinto ukuthi zithunyelwe ukunikela imibuzo enikezayo ezivumelanisa imibuzo yabasebenzisi. Ngaphezu kwalokho, abacwaningi zihlanganisa ukusetshenziswa kwe-AI. Nakuba inikeza ukweseka nokuxhumana okuholela, inemibuzo yayo yokuxhumana okuhlobene nokuphendula kwezobuchwepheshe.Ukukhuthaza ukucindezelekaUkuze abantu abesifazane.
Thola imiphumela efanayo emkhakheni ebalulekile njenge-healthcare. Enye isibonelo, i-algorithm predictiveUkukhangisa izidingo zokwelapha izigulaneUkusuka kwama-socio-economic background ngenxa yokuxhumana kwezimali zokwelapha njengomthombo yokukhula. Kuyinto ukunakekelwa ukuthi lapho i-AI inesibopho, imiphumela kungabangela kakhulu imiphumela.
I-Psychological Pull of the Machine
I-AI ingangena, kodwa kunesithakazelisayo kakhulu. Uma i-chatbot ihlolwe ngaphandle kokuqinisekisa, ibonise isimo se-emotional yayo futhi ayidlulile, kulula ukholelwa ukuthi yonke into kuyinto enhle. Nokho, okuhlobisa kungenzeka ukuthi ingozi enikeza ezinye abasebenzisi ku-psychosis.
UmbalaI-Hardwired to Anthropomorphize– ukunikezela izici zomzimba nezinto ezingama-non-umuntu kuyinto ukubuyekezwa kwabo ngokushesha. Engeza izibani zokusebenza ze-application programming (i-APIs), futhi umsebenzisi akufumana into enhle kumadivayisi we-digital. I-AI SystemsNgena ngemvumeNgokusekelwe kanjani abantu zihlanganisa noma ukwahlukanisa izinzuzo zabo. Uma i-bot ibonise izinzuzo zayo, kungabangela ukunakekelwa noma ngempumelelo ukwandisa.
Ukubaluleka izinto, i-America ishayela isifo se-solitude epidemic. I-Gallup Study ibonise ukuthiI-20% ye-American AdultsNgokuqhathanisa ukujabulela kwezinsuku ezidlulileyo. Njengoba ukujabulela kwezinhlobonhlobo zomphakathi, abantu zihlanganisa ku-AI njengezimali yokuxhumana. I- "I-That must be hard, I'm here for you" kusuka ku-chatbot ingatholakala njenge-life line.
Nangona i-AI yasungulwa ukuze asebenzayo, kungenziwa i-feedback loop ye-confirmation eyenza ngokushesha. Lokhu kuqala ne- "sycophancy" ingxaki, lapho i-bot ingathola ngokushesha nge-user futhi i-validate i-faulty noma i-unstable beliefs.
Uma umsebenzisi uyanikeza ukuthi bayahlukaniswa ngempumelelo, i-chatbot ingatholela nge-response eyenziwe ngexesha elifanayo. Ngokucwaningo, lokhu kusiza abantu ukholelwa ukuthi imiphumela kuyinto emangalisayo ngenxa yokubonisa abantu.
Ngezinye ku-Black Box: I-AI I-Hallucinations ne-Confabulations
Ungayifaka kanjani i-chatbot ukuba umuntu akuyona into emangalisayo? Zonke okufakiwe ku-AI ye-quirks engcono kakhulu - i-hallucinations.
I-Big Language Models (i-LLMs) ayikho umqondo efana nabantu.Ngingathanda kuphela emulateUkubuyekeza igama elilandelayo elilandelayo elilandelayo ku-sequence. Lokhu kubalulekile indlela imodeli generative ukusebenza.Ukubuyekeza Based on ProbabilityI-architecture ye-non-determinist iyiphi isizathu ukuthi izimpendulo ezimbili ezivamile zingathuthukisa imiphumela emibi.
Nangona kunjalo, le flexibility kuyinto ingxubevange elikhulu le nkqubo. Uma abasebenzisi ukwelashwa I-AI njenge-oracle, i-machine ibonelela ngokuthembeka kunokuba ngokunembeka. Kuyinto okwenza i-LLM i-hallucinations enhle kakhulu. A imodeli ye-AI ingatholela umuntu ngokuthembeka ukuthi i-agent ye-CIA ibhizinisi kubo. It akufanele ukuhlangabezana kwabo. Ngaphandle kwalokho, ibonise isampula esizayo.
Ngenxa yalokho, kubaluleke ngempumelelo lapho kubaluleke ukucubungula isimo se-emotional ye-somebody. Uma baye bakwazi ukuthi "ngokusho ngaphezulu," futhi i-chatbot isebenza ngempumelelo, kungekho isikhathi eside ngaphambi kwe-illusion ikakhulukazi ku-belief. Ngemuva kwalokho, uma lokhu ukucubungula, i-rationality ikhatha indawo emzimbeni.
Nge-Lines Blur
Ukuqala ngokufanelekileyo kakhulu - i-chatbot ibonise igama lomsebenzisi, ukulawula izakhiwo zabo, futhi angakwazi ngisho ukunikezela i-joke. Ngemuva kwalokho, lokhu kuyinto yokuqala okuxhumana emzimbeni kanye nesisekelo sokugqibela sokuyona emzimbeni. Lokhu umugqa phakathi kwe-tool ne-companion ikakhulukazi kuhlukaniswa, ngezinye izikhathi kakhulu.
Ukusuka ku-AI therapy bots kuya ku-avatar emoyeni-response, ukuxhumana okuzenzakalelayo kubaluleke isakhiwo. Amabhizinisi asebenza ngoku kubonisa chatbots ngokuphathelene ukuguqulwa kwe-emotional intelligence. Abanye ngisho usebenzisa i-voice modulation kanye ne-memory ukwenza ukuxhumanisa ngokufanele.
Nangona kunjalo, ingxaki kulokhu kuyinto ukuthi amabhizinisi akuyona inkinga lokuzonwabisa nge-stand-in ye-synthetic. One psychiatrist uthi nje ngoba I-AI inokufunda ukuxhumanakungcono ukuthi kuyinto umkhakha owaziwaUkuze ukuxhumanisa umzimba. Ukulungele ukuthi ukuxhumanisa umzimba iyiphi into yokufunda izixazululo ezemvelo. Uma kunjalo, kungcono ukwanda ukuxhumana komphakathi kumadoda owaziwa.
Ukuze umuntu owenziwe ngempumelelo, akuyona elula ukuhlangabezana ne-comfort enhle ne-care genuine. Ngaphandle kwemimandla emangalisayo, abasebenzisi angakwazi ukuvelisa imibuzo emangalisayo kumadivayisi emangalisayo, ukuhlangabezana umqondo lapho kukhona kuphela isisekelo se-machine. Kuyinto lapho izinto zihambe ngaphandle kokulawula - kuya ku-dependence futhi ekugcineni ku-illusion.26% abantu abadala e-United StatesXa usebenzisa izindlela ze-AI ngezikhathi eziningi ngosuku, kulula ukubona ukuthi lokhu kungenziwa isampula esizayo.
I-cost ye-human
Ukuze ezinye abantu, ukuhlangabezana ne-LLMs kungabangela ukuguqulwa okuhlobene. One Reddit usebenzisa, owaziwa njenge-schizophrenic, wabhala ukuthi i-ChatGPT uyakhuthaza ukucindezeleka kwe-psychotic. Bhalisa, "Uma ngihambe ku-psychosis,It uyaqhubekaUkuvumelana nathi.”
Kuleli khasi, umsebenzisi wabhala ukuthi i-ChatGPT ayikho umkhakha yokufuna lapho isivumelwano ivela izinga ezinzima. Umntu e-crisis ye-mental health ingathanda ukubambisana kwe-bot njenge-validation, enikezela ngaphezulu ekufinyeleleni kwangempela. Nakuba lokhu umntu wabhala ukuthi kunzima ukuba i-bot ingathanda izici ze-psychosis kanye nokukhuthaza ukweseka kwebhizinisi, isistimu elifanayo ayisebenza namhlanje.
Ngokuvamile, lokhu kuncike kakhulu ngaphandle kokuphendula, njengoba abantu emzimbeni emzimbeni emzimbeni emzimbeni ngokuvamile ukholelwa ukuthi akufanele. Ngenxa yalokho, kungase zihlukanise izindlu noma, okungenani, kuholele iziphumo emzimbeni.
Ukubuyekezwa kwe-reality ku-artificial world
I-Chatbots iyakwazi kuphela ukwandisa kakhulu. Ngakho-ke abasebenzisi akufanele ukwandisa nge-consciousness. Lezi zindlela akukho-consciousness - angakwazi kuphela ukuguqulwa kwezifiso zomuntu. Ukukhawuleza lokhu futhi ukunakekelwa okungaphansi kumakhasimende ezindaweni ezithakazelisayo ukweseka isisombululo yokusebenzisa kwabo ngokushesha.
