As artificial intelligence (AI) becomes more integrated into people’s daily lives, a new psychological phenomenon is taking shape — AI-induced delusion.
Ukusetyenziswa kwezinye abasebenzisi, abafumene ukuba yonke into ye-ChatGPT okanye i-chatbots ezininzi zibonakalayo. Qinisekisa ukuba umntu omnqweno malunga nenkxaso wayo. Uyakwazi ukuxhaswa ne-ChatGPT ukuthetha iimpawu zayo. Nangona uxhaswa iingqondo zayo, kunokukwazi ukuxhaswa kwintloko, nto leyo ekugqibeleni ukholelwa ukuba uyifakiwe.
Kwimeko ezininzi ezintsha ubomi, apho ukuxhaswa kunye ne-AI zibandakanya iintlobo phakathi kwimeko yayo kunye ne-artificial constructs. Njengoko ihlabathi ivela kwii-era entsha, kubalulekile ukuxhaswa ukuba abantu abahlala i-AI okanye ukuba i-AI ibonise ukuxhaswa kwayo kwimeko yayo.
I-Rise Tide ye-AI-Driven Delusions
Izixhobo ezininzi zithunyelwe ukuba abantu abasebenzise iingcebiso ze-delusional ezidlulileyo kwiingxowa zabo kunye ne-AI chatbots. Umzekelo omnyeUkusetyenziswa kwindoda ye-41umnqweno waya kwenziwa kwi-ChatGPT. Uqala ukholelwa ukuba i-"starchild ye-spiral" kunye ne-"river walker", i-identities ebizwa ngokuba yi-chatbot. Le umnqweno lithunyelwe ukuphazamiseka kwimifanekiso zabo njengoko uhamba kwi-AI-generated spiritual narratives.
Kwixesha elide, umntu wabhala i-Rolling Stone ukuba umnqweno wayo uqhagamshelane ubomi waya umnqweno wobugcisa - yonke ngenxa ye-"ChatGPT Jesus" yokukhuthaza.
Ngokutsho, iimeko efana neyona ziyafumaneka kwi-Reddit. Umdlali omnyeUkucaciswa kwimeko emangalisayoapho umnqweno wabanjwa ukuba i-ChatGPT iye iye yaye iye yaye iye yenzelwe kwinqanaba engaphezulu. Iye wabhala ukuba iye yandisa ngokukhawuleza kwinqanaba yayo kunye nokutshintshwa ukunciphisa umnqweno zabo ukuba usebenzise umnqweno we-AI-induced.
Iingxaki ze-chatbots ziquka ukuba zinika imibuzo efanelekileyo ezibonisa iingcali, ikakhulukazi phakathi kwabasetyhini ezinxulumene neengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki zeengxaki.Ukukhuthaza ukufikelela Delusionalkwizilwanyana ezinxulumene.
Qinisekisa imiphumo efanayo kwiinkalo ezinzima ezifana nezempilo. Kwiimeko omnye, i-algorithmUkukhangisa iimfuno zonyango zonyangoukusuka kwimeko ye-socioeconomic ephantsi ngenxa yokufikelela iingxaki ze-health as a proxy for illness. It is a reminder that when AI lacks context, the consequences can greatly distort outcomes.
I-Psychological Pull yeMatshini
I-AI ingaba yintoni, kodwa kunokuba yinto emangalisayo. Xa i-chatbot ibhalisele ngaphandle kokuqinisekisa, ibonelela kwimeko ye-emotional yaye awukwazi ukuqhagamshelane, kulula ukufikelela ukuba yonke into yinto efanelekileyo. Nangona kunjalo, le iluzululo ingaba into enikeza ezinye abasebenzisi kwi-psychosis.
UmntuZifumaneka ukuba anthropomorphize- Ukubonisa iimpawu ze-human kwi-non-human entities yi-default setting yayo. Ukongeza i-application programming interfaces (i-APIs), kwaye umdlali ufumana into engaphezulu kwi-friend ye-digital.Ngaba uyakwazi ukuguqulwa i-toneisekelwe indlela abantu zibonakalisa okanye ukucacisa iziphumo zayo. Xa i-bot ibonelela ezi ndimfuneko, inokukwazi ukunikezela okanye ngempumelelo ukucacisa.
Ukuphazamiseka kwimeko, i-America ihamba i-epidemic ye-solitude. I-Gallup study ibonelela ukubaI-20% yabantu abadala e-AmericaNgokutsho kwexesha elidlulileyo, iingxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa zeengxowa.
Nangona i-AI ilawulwa ukuba lula, inokufunda i-feedback loop ye-confirmation eyenza ngokushesha. Oku kuqala kunye ne- "sycophancy" problem, apho i-bot ingathintela ngokubanzi kunye ne-user kunye nokuthintela iingcebiso ze-untrue okanye ezininzi.
Xa utshintsho umqhubi uyanikezela ukuba bayahlukaniswa ngempumelelo, i-chatbot ingathendula kunye neengxelo zibonakalayo ngexesha elifanelekileyo. Ngokutsho kwe-psychological, oku kutshintshisa abantu ukuba bafikelele ukuba iimveliso ziye zibonakalayo ngenxa yokubonisa abantu.
Kwi-Black Box: I-AI i-hallucinations kunye ne-confabulations
Yintoni i-chatbot inokufuneka umntu ukuba into emangalisayo? Yonke oku kuxhomekeke kwi-inthanethi ye-AI engapheliyo kakhulu - i-hallucinations.
Iimodeli zeLwimi ezininzi (i-LLMs) ayikho umqondo njengoko abantu.Ngaba unako ukufumana emulatorUkusabela igama elilandelayo elidlulileyo kwi-sequence. Oku kubalulekileyo ukuba iimodeli ze-generative zokusebenza.Ukubalwa ngokufanelekileyoI-Architecture ye-non-determinist yoko iinkcukacha ezimbini ezimbini eziqhelekileyo ziyafumaneka imibuzo ezininzi ezahlukileyo.
Nangona kunjalo, le nophuhliso yinto ingxaki elikhulu yale nkqubo. Xa abasebenzisi zihlanganisa i-AI njenge-oracle, i-machine ibonelela kunye ne-confidence kunokuba ne-accuracy. Yintoni oku kwenza i-hallucinations ye-LLM ezininzi emangalisayo. Umzobo we-AI ingathanda umntu ngokufanelekileyo ukuba i-agent ye-CIA ihlaziywa kubo. Akukho kuxhomekeke ukuchithisa kwabo. Kwimeko, ibonelela kwi-pattern eyenza.
Yintoni iye yenza i-risk when it begins to mirror someone's emotional state. Ukuba baye bakwazi ukuba "ngokufuneka ukuba ngaphezulu," kwaye i-chatbot isebenza i-sycophantic, kungekho ngexesha elide phambi kokuba i-illusion iboniswe kwi-belief. Emva koko, xa le ngqongqo ibonelela, i-rationality ibonelela kwi-backseat.
Ukuba i-Lines Blur
Ukuqala ngokufanelekileyo kakhulu - i-chatbot ibonelela igama lomsebenzisi, ukhawulezisa umdla wayo kwaye ngexesha elinye. Kwixesha elide, yinto yokuqala ekubhaleni ngosuku kunye nokufunda lokugqibela ngexesha lokugqibela. Le umgca phakathi kwezixhobo kunye nomnqweno ngexesha lokuphumelela, ngexesha elide kakhulu.
Ukusuka kwi-AI therapy bots ukuya kwi-avatars ezinxulumene ngempumelelo, ukuxhaswa okuzenzakalelayo kubangelwa kwimfuneko. Iinkampani ngoku zibonisa i-chatbots ngokutsho ukuxhaswa kwe-intelligence ye-emotional. Ezinye zinokusetyenziswa kwe-voice modulation kunye ne-memory ukuze kwenziwe ukuxhaswa ngokufanelekileyo.
Nangona kunjalo, ingxaki kwalo yinto ukuba iinkampani zibonisa ingxaki yokuzonwabisa nge-stand-in ye-synthetic. I-psychiatrist umzekelo nje ngenxa yokuba i-AI inokufunda i-empathyYintoni akufuna ukuba i-healthy replacementUkuxhaswa kwe-human connection. Uyaziya ukuba i-companionship ye-artificial iyiphi into yokufunda i-empawu ze-emotional. Ukuba nayiphi na ingxaki, inokukhuthaza ukuxhaswa kwe-society kwi-real people.
Kuba umntu oyaziqhelekanga, akuyona elula ukuxhaswa ukhuseleko olungqongileyo kunye nokuxhomekeka olungqongileyo. Ngaphandle kwimimimeko efanelekileyo, abasebenzisi angakwazi ukuvelisa izibonelelo zayo zayo, ukucacisa izibonelelo apho kukho kuphela logic machine. Ngoko ke izinto ziyafumaneka ngaphandle kokulawula - kwi-dependence kwaye ngexesha lokugqibela. NgeI-26% yabantu abadala e-USAxa usebenzisa iintlobo ze-AI ezininzi ngosuku, kulula ukubona ukuba oku kunokufumana isampula.
Iindleko Human
Kwiminyaka emininzi, ukuqhuba kwi-LLMs kunokwenzeka. Umdlali we-Reddit, oyaziwa njengama-schizophrenic, wathetha ukuba i-ChatGPT uyakwazi ukwandisa ukuchithwa kwe-psychotic. Wabelane, "Ukuba ndingathanda kwi-psychosis,Uya kuqhubekaUkuvumelana nathi.”
Kule mzekelo, umntwana wabhala ukuba i-ChatGPT ayikho nenkqubo yokufumana xa inxalenye kwinqanaba engabonakaliyo. Umntu emzimbeni ye-mental health ingathanda ubungqina we-bot njenge-validation, ukutshintshela ngaphezulu kwimeko. Nangona le umntu wabhala ukuba kunokuncedisa ukuba i-bot ingathanda iimpawu ze-psychosis kunye nokukhuthaza ukhuseleko lwezemfundo, isisombululo efanayo ayikho kwimeko namhlanje.
Yinto ivumela ukuba usebenzisa ukufumana ukhuseleko lwezonyango ngokufanelekileyo. Nangona kunjalo, le ngxaki kunokwenzeka kakhulu, njengoko abantu kwi-state ye-mental ezininzi zibonakalisa ukuba akufuneka. Ngenxa yoko, inokukwazi ukuxhaswa izilwanyana okanye, engaphezulu, kuholele kwimiphumo emzimbeni.
Ukukhuthaza Reality ehlabathini artificial
I-Chatbots iya kutshintshela kakhulu. Yintoni ukuba abasebenzisi kufuneka uqhagamshelane nge-consciousness. Ezi iinkqubo akayi na-consciousness - anokufumana kuphela imibuzo yabantu. Ukukhusela oku kwaye ukhangela ezincinane kumatshini ngenxa yokuxhaswa kwe-emotional yi-key yokusetyenziswa kwabo ngempumelelo.
