Sekungamashumi eminyaka siphupha ngokuba njengoDokotela Dolittle, lowo mlingiswa okwaziyo uxoxe nanoma yisiphi isilwane Kanjalo nje. Lokho okwakuyizincwadi zezingane kanye nama-movie aseHollywood manje sekuqala ukusondela eqinisweni ngenxa yobuhlakani bokwenziwa (AI). Asikafiki ezingeni lokubuza inja yethu ukuthi yayicabangani ngokuhamba noma ukucela umkhomo ukuba usitshele indaba yokuphila kwawo, kodwa isayensi ihamba ngokushesha kakhulu kulolo hlangothi.
Eminyakeni yamuva nje, amaphrojekthi ocwaningo asatshalaliswe emhlabeni wonke aqale ukusebenzisa ama-algorithms athuthukile ukuze hlaziya imisindo, ukushukuma komzimba kanye nokubonakaliswa kwezinhlobo eziningiKusukela ezinkukhu, ezingulubeni, nasezinjeni kuya emikhomeni, ezihlengethwaneni, ezindlovu, nasezinkalweni, la maphrojekthi afuna okuthile okujulile kakhulu kunomhumushi wezilwane ezifuywayo olula: ahlose ukuqonda ngempela ukuthi izilwane zixhumana kanjani, zithuthukise inhlalakahle yazo, futhi, zishintshe ubudlelwano bethu nazo kanye nezindawo esizizungezile.
Kusukela emcabangweni kaDokotela Dolittle kuya emalabhorethri e-AI
Inganekwane yokukwazi ukukhuluma nezilwane zasendle iyathandeka ngoba ixhumeka nesifiso somuntu: ukuqonda ukuthi izilwane zizizwa kanjani futhi zicabangani ezisizungezile. Izincwadi sezivele zihlole lokhu noDokotela Dolittle, udokotela owayeka ukwelapha abantu ukuze bazinikezele ezilwaneni lapho ethola ukuthi uyakwazi ukukhuluma ulimi lwabo, waba yisazi semvelo nomhloli. Kwakubonakala sengathi kuyimilingo. Namuhla, isofthiwe isiqala ukuthatha indima "yomhumushi."
Umehluko omkhulu kusukela esikhathini esidlule ukuthi manje sesinayo amamodeli okufunda okujulile akwazi ukucubungula izigidi zamaphuzu edatha umsindo, isithombe, kanye nomongo ngesikhathi esifushane kakhulu. Lokho okwakuyimibono yezazi ze-ethology ezihlome ngama-notebook asensimini kanye nama-tape recorder manje sekuba yizizindalwazi ezinkulu zedijithali kanye namanethiwekhi ezinzwa afuna amaphethini angabonakali endlebeni nasesweni lomuntu.
Ubuhlakani bokwenziwa abuqondi izilwane ngendlela umuntu okhuluma izilimi ezimbili angaziqonda ngayo, kodwa bungakwazi bona ukujwayela ekukhulumeni kwabo, ukunyakaza kanye nezinkulumo zaboukuzihlobanisa nezimo ezithile futhi, ngolwazi olwanele, ziphakamise izincazelo ezingaba khona. Luhlobo lwe-Rosetta Stone olusebenzisa i-algorithmic olusetshenziswa embusweni wezilwane.
Ngesikhathi esifanayo, ososayensi ngokwabo baxwayisa ngokuthi le ntshiseko kumele ihambisane nokuqapha: Ukuhumusha akukhona nje ukuguqula imisindo ibe amagamaKuhilela futhi ukuqonda ithoni, ukuhlekisa, inhloso, kanye nomongo, into eyinkimbinkimbi ngisho naphakathi kwabantu. Lobu bunzima buyanda uma kukhulunywa ngezinye izinhlobo.
Iphrojekthi Yezinhlobo Zomhlaba: ukuhlolwa okuhle kokuqonda ulimi lwezilwane
Phakathi kwezinhlelo ezibaluleke kakhulu yi-Earth Species Project (ESP), inhlangano engenzi nzuzo eyazalelwa eSilicon Valley enomgomo, ephepheni, ozwakala ungenakwenzeka: funda izinhlelo zokuxhumana izinhlobo eziningi ngangokunokwenzeka emhlabeni futhi udale amathuluzi e-AI angakwazi ukuwachaza kuphela, kodwa futhi nokukhiqiza imiyalezo eqondakalayo kulezo zilwane.
I-ESP ihlanganisa onjiniyela, ochwepheshe bokufunda komshini, kanye nochwepheshe bokuziphatha kwezilwane. Indlela yabo ihilela ukuqoqa imiqulu emikhulu yokuqoshwa komsindo, amavidiyo, kanye nedatha yezemvelo ezihlotshaniswa nezimo ezithile kakhulu: ukondla, ukudlala, ukuzala, ingozi, ukuxhumana nabantu… Ngalolu lwazi baqeqesha amamodeli afuna “amayunithi amancane encazelo” ezingomeni, ekukhaleni, ekuchofozeni noma ekubhongeni kohlobo ngalunye.
Lo msebenzi uthole ukubonakala kwabezindaba ngo-2023, lapho isazi sezinto eziphilayo zasolwandle uMichelle Fournet, ebambisene nobuchwepheshe be-ESP ecosystem, ephumelela chaza uhlobo lokukhala komkhomo we-humpback esimweni sokuxhumana nabantu. Nakuba kwakungeyona "inguqulo yegama negama," kwakuwubufakazi bokuthi i-AI ingasiza ekuhlukaniseni amaphethini okuxhumana ayengabonakali ngaphambilini.
Abantu abaphethe i-Earth Species Project basebenza ngezinhloso ezinhle zesikhathi: cishe eminyakeni engamashumi amathathu bafisa ukukwazi ukuhumusha, okungenani ngokwengxenye, imisindo yezinhlobonhlobo ezinkulu zezilwaneBaze bakhulume nangesikhathi esizayo lapho amadokhumentari emvelo azohlanganisa khona imibhalo engezansi ekhiqizwa ngokuzenzakalelayo ngemisindo yangempela yezinhlobo eziqoshwe, okunciphisa isidingo sabalandisi abangabantu.
Ukusekela lokhu kukhuphuka, i-ESP isanda kuthola cishe Imali engu-$17 millionLokhu kuzobavumela ukuthi baqashe onjiniyela abengeziwe, baphinde kabili ubukhulu bethimba lamanje, futhi basheshise ukuthuthukiswa kwamamodeli afana ne-NatureLM-audio, egxile kwi-bioacoustics futhi yakhelwe ukufunda ngqo kumasignali emvelo.
I-NatureLM-audio kanye "neRosetta Stone" yokucula
Enye yentuthuko ephawulekayo ehlobene nalolu hlobo lomsebenzi yi-NatureLM-audio, imodeli ye-AI eqeqeshwe nge ukuqoshwa okukhulu kwemisindo yezilwane zasendleUmsebenzi wayo oyinhloko, okwamanje, akukhona ukuhumusha "imishwana" ephelele, kodwa ukuhlonza ukuthi ubani okhipha umsindo, kusiphi isimo futhi yiziphi izinguquko ezihlelekile.
Ekuhlolweni okwenziwe ngezinhlobo ezahlukene, i-NatureLM-audio ifinyelele ukuhlukanisa ngokunemba okunengqondo ubulili, iminyaka ecishe ifane kanye nenani labantu okukhona ekuqopheni. Lokhu, okungase kubonakale njengemininingwane yobuchwepheshe, kuyisihluthulelo samaphrojekthi okulondolozwa kwemvelo: kusenza sazi, isibonelo, ukuthi inani lezilwane ezincelisayo zasolwandle lakhiwa yizingane eziningi ezincane, ukuthi kunezinsikazi ezizalayo, noma ukuthi kwenzeka ukunyakaza okungavamile.
Uma kwenzeka izilwane ezisengozini yokuqothulwa njenge-beluga noma ezinye izilwane ezincelisayo, abacwaningi bakholelwa ukuthi lalela ngokuhlelekile "izingcingo" zabo futhi uziqonde kangcono Ingasetshenziswa ukuvimbela ukushayisana nemikhumbi, ukunciphisa umsindo ezindaweni ezibucayi, noma ukubikezela izinguquko emizileni yokufuduka ngaphambi kokuba iguqulelwe ku- imicu noma ukufa okuphezulu.
Le ndlela "yokulalela ukuze wonge" iphawula ushintsho oluyisisekelo: esikhundleni sokufunda izilwane ngaphandle kuphela, inhloso iwukuba... bahlanganise ukuxhumana kwabo njengomthombo wedatha obalulekile ukwenza izinqumo mayelana nokuphathwa nokuvikelwa kwemvelo.
Izindlovu ezibizana ngamagama kanye nezinye izinto ezitholakele ezimangazayo
Enye yezinto ezitholwe ngendlela exakile eminyakeni yamuva nje ivela ocwaningweni lwezindlovu zase-Afrika. Ngokuhlaziya kakhulu imisindo yazo, ososayensi bathole ukuthi, ezimweni ezithile, umsindo othize wawuqondiswa njalo kumuntu oyedwa., okukodwa kwe amalungu eqembungubani owayephendula. Lapho lowo msindo ushintsha, indlovu ehlukile yayiphendula.
Konke kusikisela ukuthi la ma-colossi angasebenzisa uhlobo "lwamagama afanele", into efana amalebula omsindo ngamanye ukubhekana. Lo mehluko ucashile kangangokuthi indlebe yomuntu ayikwazi ukuwuhlukanisa, kodwa ama-algorithms ayakwazi ukubona amaphethini ahlelekile asikisela isakhiwo solimi esiyinkimbinkimbi kakhulu kunalokho okwakucatshangwa ngaphambili.
Olwandle, amaphrojekthi afana ne-CETI (edlala ngendlela efanayo ne-SETI, uhlelo olufuna ubuhlakani bangaphandle komhlaba) agxile emikhomeni yesidoda. Umsebenzi wawo uwukuba ukuphula ukuchofoza kanye namaphethini anesigqi akhishwa yilezi zilwane, zihlobanise nezimo ezibonwe futhi, ngesikhathi kanye nedatha, ziphethe incazelo yazo ecishe ifane.
Uma i-CETI ikwazi ukuqhubekela phambili kulokhu, izovula indlela eya ukucacisa izilimi zezinye izilwane ezincelisayo zasolwandleKodwa-ke, uhlobo ngalunye luzodinga isu lalo kanye namamodeli alo athile. Akukho "umhumushi wendawo yonke" oyedwa osebenzela yonke imfuyo, okungenani okwamanje.
Eduze kobuso kanye nokuphila kwethu kwansuku zonke, kuye kwenziwa intuthuko ephawulekayo ngezilwane zasemapulazini: izingulube, izinkukhu kanye nezinkawu ezikhungweni zezilwane Baqashwa ngemishini yomsindo nevidiyo enesinqumo esiphezulu ukuqeqesha i-AI ukuqaphela imizwa, amazinga okucindezeleka, noma ukungakhululeki.
Izinkukhu, izingulube nezinkawu: I-AI ingena emapulazini nasezindaweni zoo
ECanada, iqembu labacwaningi eHalifax lisebenza nemihlambi yezinkukhu ezihlaziya ukuklukha kwazo kusetshenziswa ama-spectrogram kanye namamodeli okuhlukaniswa. Umqondo ulula kodwa unamandla: ukuxhumanisa uhlobo ngalunye lokubiza nesimo sengqondo kanye nesimo esithilenjengokuzizwa uzolile, ucindezelekile, ugula, noma ungakhululekile ngenxa yokuminyana.
Uma i-AI ingaqaphela ngokwethembeka ukuthi umhlambi wezinyoni uyahlupheka, abalimi bangakwazi Lungisa umoya ongenisa umoya, ubuningi, ukunikezwa kwamandla, noma ukukhanya. isikhathi eside ngaphambi kokuba kuvele izimpawu zangaphandle ezisobala, kunciphisa ukuhlupheka futhi kuthuthukise umkhiqizo ngaphandle kokulinda ukubona ukungabikho noma ukulimala.
EYurophu, amanye amaqembu asebenzisa amanethiwekhi ezinzwa ukuhlaziya imisindo kanye nezimpawu zomzimba zezingulube kanye nezinhlobo ezahlukene zezinkawu ezikhungweni zezilwane. Ngenxa yalokhu, ama-algorithms angakwazi hlukanisa amagama ngokwezigaba ezifana nokudlala, ukwesaba, ulaka, noma ukuzolaukunikeza abanakekeli "isixwayiso kusenesikhathi" uma kwenzeka kuba nengxabano yomphakathi noma ukucindezeleka okungapheli.
Lezi ntuthuko zisikisela ukuthi, maduze nje, amapulazi amaningi emfuyo angafaka izinhlelo zokulalela nokuhlaziya okuzenzakalelayo abaxwayisa ngezinkinga zenhlalakahle yezilwane amahora angama-24 ngosuku, izinsuku eziyi-7 ngesonto, ngaphandle kokulinda ukuvakashelwa njalo ngudokotela wezilwane noma ukufa kukhuphuke kakhulu.
Akukhona nje ukukhiqiza okwengeziwe, kodwa ukuthatha ngokungathi sína lokho izilwane ebezisitshela kona isikhathi eside ngemisindo yabo kanye nokuziphatha kwabo, manje okukhuliswe futhi kuhlelwe ngamathuluzi e-AI.
Izinja namakati: lapho i-AI ingena ngokunyenya ekhaya
Uma kunezilwane abantu abaphuphe ngazo ukukhuluma nazo izizukulwane ngezizukulwane, yizinja namakati. Ukuhlalisana kwansuku zonke kuqinisa umbono wokuthi “Bayasiqonda” kakhulu kunalokho abangakushoI-AI isiqala ukufaka izinombolo kulowo mbono.
E-University of Michigan, ngokubambisana ne-INAOE eMexico, ithimba lavumelanisa imodeli ye-Wav2Vec2 eyaziwayo—eyayiklanyelwe ukuqashelwa kwezwi lomuntu—nezwe lezinja. Baqeqesha i-AI ngokuqopha Izinja ezingu-74 ezimweni ezilawulwayo ngokomzwelo (injabulo, ukwesaba, ukukhungatheka, ukudlala…) futhi wamcela ukuthi ahlukanise amagama ahlukene.
Uhlelo lukwazile ukubona, ngokunemba okucishe kube okuphelele 70%isimo semizwa yesilwane, kanye nezimo eziguquguqukayo njengohlobo, ubulili, kanye nobudala. Lesi yisikhathi sokuqala imodeli eyenzelwe ulimi lwabantu isetshenziswa ngempumelelo ekubhongeni, okubonisa ukuthi Kunesakhiwo esanele kulezo misindo ukuze sihunyushwe ngokwe-algorithm.
Ngamakati, iphrojekthi yocwaningo eNyuvesi yaseMilan iqophe ama-meow ezimweni ezahlukene zansuku zonke—elinde ukudla, ngesikhathi sokuxubha, ngezikhathi zokuzihlukanisa—futhi yaqeqesha i-algorithm yokwenza qaphela isimo esithile esihlotshaniswa nohlobo ngalunye lwe-meowImiphumela ibonise izinga lokunemba elimangalisayo elingu-96%.
Lolu hlobo locwaningo luholele ekuthuthukisweni kwe-MeowTalk, uhlelo lokusebenza olwasungulwa ngo-2020 nguJavier Sánchez, owayengunjiniyela we-Amazon Alexa. Uhlelo seluvele lucubungule okungaphezu kwalokho ama-meow ayizigidi eziyinkulungwane futhi iyakwazi ukuhlukanisa izinhloso zokuxhumana ezifika kweziyisishiyagalolunye, njengokuthi "Ngifuna ukudlala", "Ngilambile" noma "ngiyeke ngedwa", futhi ilungise iphrofayili yekati ngalinye.
Ubuso obukhulumayo: I-AI efunda izinkulumo zezilwane
Akukhona nje kuphela ngezwi. Abacwaningi bayanda ngokuhlanganisa umsindo nombono wekhompyutha ukuze bahlaziye ukubonakaliswa okuncane kobuso kanye nokuma komzimba ezinhlotsheni ezahlukene. Emakatini, isibonelo, lokho okubizwa ngokuthi i-Feline Grimace Scale sekudumile, okulinganisa izinguquko ezincane ezindlebeni, emehlweni, emadevuni nasempumulweni ukuze kuhlolwe ukuthi isilwane sizwa ubuhlungu yini.
Izinhlelo zokusebenza ezifana ne-Table zisebenzisa i-AI ukuze bamba futhi ucubungule izithombe zobuso bekati futhi ukhiphe isilinganiso samathuba obuhlungu. EJapane, uhlelo lwe-CatsMe! lufinyelele ukunemba okungaphezu kuka-95% ekutholeni ngokuzenzakalelayo ukuhlupheka ngo-2024, ngemuva kokuqeqeshwa ezithombeni eziyizinkulungwane ezibhalwe odokotela bezilwane.
Ngesikhathi esifanayo, izifundo zokuqhathanisa zikhombisile ukuthi abantu nezinja babelana ngesilinganiso 38% wesimo sobuso esibonakalayoLesi sibalo sikhuphukela cishe ku-34% kumakati. Lokhu kusikisela ukuthi kunesisekelo esifanayo lapho i-AI ingasebenza khona ukuhumusha ukushukuma komzimba kube yizimo eziyisisekelo zemizwa ngokuthembeka okwandayo.
Umthelela ongokoqobo mkhulu kakhulu: imitholampilo yezilwane, izindawo zokukhosela kanye nezinhlangano zenhlalakahle yezilwane zingakwazi beka phambili ukunakekelwa kwalezo zilwane uhlelo oluzibona njengezisezinhlungwini ezinkulungisho noma zibonakala zilungile ekuqaleni. Ngakho-ke i-AI isebenza njenge-"glass yokukhulisa ngokomzwelo" ephelelisa—engalokothi ithathe indawo—yokwahlulela kobungcweti.
Ngale kwempilo, lokhu kufundwa okuzenzakalelayo kwezinkulumo kuvula umnyango amathoyizi, amakhamera asekhaya kanye nemigexo ehlakaniphile ezitshela umnakekeli ukuthi isilwane sakhe sinesithukuthezi, sikhathazekile noma simatasa kakhulu, okusiza ekulungiseni imikhuba yaso futhi kucebise indawo esizungezile.
Abahumushi, ama-avatar akhulumayo, kanye nezinhlelo zokusebenza zokukhuluma nesilwane sakho sasekhaya
Njengoba ucwaningo luqhubeka, imikhiqizo yezentengiselwano iqala ukubonakala, ethembele kulezi ntuthuko zesayensi, inikeza ukuhumusha okulinganiselwe noma imibono ehlekisayo yalokho izinja namakati “okungakusho”Lapha umgomo awuseyona nje isayensi, kodwa futhi ukuzijabulisa nokuthuthukisa isibopho somuntu nesilwane.
Iphrojekthi Yezinhlobo Zomhlaba ngokwayo isebenza kuhlelo lokusebenza lwe- Ukuhumusha okusekelwe ku-AI kwezinja namakatiokungasebenza ngendlela efanayo ne-chatbot: umsebenzisi ufaka noma aqophe amasignali (ukukhonkotha, ukumemeza, ukunyakaza), uhlelo luyawahlaziya maqondana nesizindalwazi salo futhi lukhiqize impendulo yombhalo emele inhloso engenzeka kakhulu.
Ngesikhathi esifanayo, izixazululo ezifana Ama-Petpuls (ikhola elihlaziya imizwa yezinja)Amapulatifomu ezinkinobho zohlobo lwe-FluentPet—avumela isilwane ukuthi sicindezele izithonjana ngamagama aqoshiwe ngaphambilini—noma amaphrojekthi afana ne-Zoolingua, ahlose ukuhumusha imisindo, ukushukuma kanye nezinkulumo ngendlela ehlukahlukene.
Enye indawo ethole ukuthandwa kakhulu yi-generator. amavidiyo ezilwane ezifuywayo ezikhulumayoLawa mathuluzi akuvumela ukuthi ulayishe isithombe senja noma ikati, ufake umbhalo othile, bese uvumela i-AI ukuthi ivumelanise umlomo wayo kanye nokubonakaliswa kwayo ukuze ibukeke sengathi isilwane siyakhuluma. Nakuba kuwumsebenzi wobuhle kuphela, kusiza ukujwayela umphakathi jikelele ngombono wokuthi ngolunye usuku lokho kunyakaza kwezindebe kungahlotshaniswa nomyalezo osekelwe kudatha yangempela.
Eziningi zalezi zinhlelo zokusebenza zinikeza izinhlelo zamahhala ezinokusetshenziswa okulinganiselwe kanye namamodeli okubhalisela ukufinyelela izici ezithuthukisiwe noma isizukulwane esingenamkhawulo, okubonisa imakethe ekhulayo ezungeze "Ukuhumusha ngokomzwelo" kwezilwane ezifuywayo.
Kusukela epulazini kuya esibhedlela sezilwane: ukusetshenziswa okusebenzayo kanye nezinyathelo zokuphepha
Ngaphandle kwezokuzijabulisa, ukuhlakanipha okwenziwayo okusetshenziswa olimini lwezilwane kubonisa isithembiso. ukuguqula ngokujulile umkhuba wezilwane kanye nokuphathwa kwezilwane zasekhaya nezasendle.Ukuthola ubuhlungu, ukwesaba, noma ukukhathazeka kusenesikhathi kungenza umehluko phakathi kokwelashwa ngesikhathi esifanele nesimo esingenakuguqulwa.
Ezimweni zezokwelapha, izinhlelo zokuhlaziya umsindo nevidiyo zingahlanganiswa emakamelweni okulinda, ezindaweni zokuhlinza, noma emawodini ezibhedlela ukuze qapha izinguquko ekukhulumeni nasekukhulumeni ezingabonwa yiso lomuntu, okubangela izexwayiso lapho into ingahambisani nesimo senduduzo.
Emvelweni yemvelo, ubuchwepheshe obufanayo bungasiza abaqaphi bamapaki kanye nabazilondolozi ukuthi thola amaphethini okucindezeleka ezinhlotsheni zezindlovu, obhejane noma izinyoni zasolwandleLungiselela imizila yokuqapha futhi ulindele izingxabano nemisebenzi yabantu njengokuhamba ngemikhumbi, ezolimo, noma ukwakhiwa.
Kodwa-ke, izinhlangano zobungcweti ezifana neCanadian Veterinary Medical Association zigcizelela isidingo se- ukuqaphela okukhulu ekusebenziseni la mathuluziBagcizelela ukuthi i-AI isesencane, inobandlululo namaphutha, futhi ayinakukwazi ukuthatha indawo yokuhlolwa kwezokwelapha, ulwazi, kanye nokwahlulela kochwepheshe.
Isincomo samanje sicacile: sebenzisa i-AI njenge ukwesekwa okwengeziweAkukaze kube yinto ethatha indawo, futhi kuqeqeshwe kahle ekusetshenzisweni kwayo. Kungenzeka ukuthi odokotela bezilwane besikhathi esizayo bazofaka izifundo ekuhlaziyweni kwedatha nasekuphathweni kwe-algorithm ekuqeqeshweni kwabo, babe uhlobo "lodokotela beDolittle" banamuhla, kodwa bahlale benamathele ngokuqinile esisekelweni sesayensi.
Umthelela wokuziphatha kanye nowenhlalo wokukwazi "ukulalela" izilwane
Uma sifika eqophelweni lapho i-AI ivumela izilwane ukuthi ziveze ngokucacile lokho ezikuthandayo, ubuhlungu bazo, noma ukwenqaba kwazo imikhuba ethile, Imiphumela yokuziphatha izoba mikhulu kakhuluAsisakwazi ukwenza sengathi asazi ukuthi bazizwa kanjani ngezindlela ezithile zokuxhashazwa, ukuthuthwa, noma ukuvalelwa.
Emkhakheni wezilwane, isibonelo, ukuxhumana okucacile kungaphoqa cabanga kabusha ngobukhulu bezikheji, ubuningi babantu, noma izindlela zokuhlabaEmbonini yezokuzijabulisa, amasekisi noma imibukiso enezilwane ingabona ukufaneleka kwazo kungatshazwa nakakhulu uma amazinga aphezulu okucindezeleka eboniswa, abikwe ngqo ngabadlali ngokwabo.
Kukhona futhi nezingozi zokusetshenziswa kabi kwalezi zobuchwepheshe. Uhlelo olufanayo olungavumela xwayisa imikhomo ukuba isuke endaweni enamathele kuyo Lokhu kungasetshenziswa ngabadobi bemikhomo abangathembekile ukuze bayenge ezindaweni zokubamba noma zokuzingela. Yingakho ochwepheshe abaningi befuna izinhlaka zomthetho ezicacile kanye nemithetho yokuziphatha eqinile.
Enye inkinga ebucayi yilena yokwakheka komuntu: isilingo sokwenza ukuveza imizwa nemicabango yabantu ezimpawini zezilwane okungase kungasho ngqo lokho esikucabangayo. Lapha, i-AI ingayenza inkinga ibe yimbi kakhulu uma ihumusha imiyalezo ibe ulimi lwemvelo oluzwakala "lungumuntu", idale umbono wengxoxo empeleni eyimodeli yezibalo.
Noma kunjalo, ucwaningo oluningi luyavuma ukuthi ukunikeza "izwi" kwezinye izinhlobo kungenza siqinise uzwela lwethu ngabo kanye nezinhlelo zemveloukusiza ekulweni nokushintsha kwesimo sezulu kanye nokulahlekelwa yizinhlobonhlobo zezinto eziphilayo ngobudlelwano obuqondile nezinye izinto eziphilayo.
Konke kusikisela ukuthi, ngokuhamba kwesikhathi, sizohlala ndawonye nezinhlelo ze-AI ezihlanganisa idatha yomsindo, ividiyo, kanye neyomzimba kumamodeli e-multimodal, akwazi ukukhiqiza ukuhumusha okuqondiswayo okufanelana "nesilulumagama" somuntu ngamunye, hhayi ukuthatha indawo yobuhlobo bethu obuqondile, kodwa ukubucebisa.
Ithuba langempela lokuthi i-algorithm ingasitshela ngokwethembeka okunengqondo lokho inja, ikati, indlovu, noma umkhomo ozama ukukuveza kuzoshintsha indlela esibabona ngayo, sibanakekele, futhi sinqume ukuthi yini eyamukelekayo esingayenza ngabo, okusisondeza kancane kulelo zwe lamaphupho lapho zonke izidalwa ekugcineni zinento ezingasitshela yona futhi thina, okokuqala, sinamathuluzi abalulekile okuzilalela ngempela.
