Ubuchopho bokwenziwa: umcabango wokuthakatha emshinini
I-Artificial intelligence akufanele ibe ikhophi yobuhlakani bomuntu, ngakho-ke iphrojekthi yokudala ubuchopho bokwenziwa, ikhophi yezobuchwepheshe yomuntu, iyindawo ehluke kancane yocwaningo. Kodwa-ke, kungenzeka ukuthi kwesinye isigaba sokuthuthuka le phrojekthi ingahlangana nokuthuthukiswa kwe-AI. Kwangathi lo kungaba umhlangano oyimpumelelo.
I-European Human Brain Project yethulwa ngo-2013. Akuchazwa ngokusemthethweni ngokuthi "iphrojekthi yobuchopho bokwenziwa". Kunalokho, igcizelela isici somqondo, isifiso sokubonisa kangcono isikhungo sethu somyalo. Amandla okuqamba amanga e-WBP awanalo ngaphandle kokubaluleka njengesikhuthazo sokuthuthukiswa kwesayensi. Kodwa-ke, akunakuphikwa ukuthi umgomo wososayensi abasebenza kule phrojekthi ukudala ukulinganisa kobuchopho obusebenzayo, futhi lokhu kungakapheli iminyaka eyishumi, okungukuthi, kusukela ku-2013 kuya ku-2023.
Ososayensi bakholelwa ukuthi imephu yobuchopho enemininingwane ingaba usizo ekudaleni kabusha ubuchopho bomuntu. Ikhulu lezigidigidi ezixhunywe kuyo zakha yonke into evaliwe - ngakho-ke, umsebenzi omkhulu uyaqhubeka wokwakha imephu yalokhu bunzima obungacabangeki, obubizwa ngokuthi i-connectome.
Leli gama laqala ukusetshenziswa emaphepheni esayensi ngo-2005, ngokuzimela ngababhali ababili: u-Olaf Sporns waseNyuvesi yase-Indiana kanye no-Patrick Hagmann we-University Hospital of Lausanne.
Ososayensi bakholelwa ukuthi uma bebala yonke into eyenzekayo ebuchosheni, kuzokwazi ukwakha ubuchopho bokwenziwa, njengomuntu, bese, ngubani owaziyo, mhlawumbe kangcono ... Iphrojekthi yokwakha i-connectome egameni kanye nengqikithi ibhekisela kuphrojekthi eyaziwa kakhulu yokucacisa i-genome yomuntu - i-Human Genome Project. Esikhundleni somqondo we-genome, iphrojekthi eqalisiwe isebenzisa umqondo we-connectome ukuchaza ingqikithi yokuxhumana kwe-neural ebuchosheni. Ososayensi bathemba ukuthi ukwakhiwa kwebalazwe eliphelele lokuxhumana kwe-neural ngeke lithole isicelo hhayi ekusebenzeni kwesayensi kuphela, kodwa nasekwelapheni izifo.
www.humanconnectomeproject.org
Eyokuqala futhi kuze kube manje ukuphela kwe-connectome eyaziwa ngokugcwele inethiwekhi yokuxhumana kwe-neuronal ohlelweni lwezinzwa lwe-caenorhabditis elegans. Yasungulwa ngokwakhiwa kabusha kwe-1986D kwesakhiwo sezinzwa kusetshenziswa i-electron microscopy. Umphumela womsebenzi wanyatheliswa ngo-30. Njengamanje, iphrojekthi enkulu yocwaningo eyenziwa ngaphakathi kohlaka lwesayensi entsha ebizwa ngokuthi i-connectomics yi-Human Connectome Project, exhaswe yi-American National Institutes of Health (inani eliphelele lamaRandi ayizigidi ezingu-XNUMX).
Intelligence Algorithm
Ukwakha ikhophi yokwenziwa kwengqondo yomuntu akuwona umsebenzi olula. Kungase kube lula ukuthola ukuthi ubuhlakani bomuntu buwumphumela we-algorithm elula echazwe kumagazini kaNovemba 2016 weFrontiers in Systems Neuroscience. Itholwe nguJoe Tsien, isazi sezinzwa e-Augusta University of Georgia.
Ucwaningo lwakhe lwalusekelwe kulokho okubizwa ngokuthi i-theory of connectionism, noma ithiyori yokufunda enkathini yedijithali. Isekelwe enkolelweni yokuthi inhloso yokufunda iwukufunda ukucabanga, okuza kuqala kunokuthola ulwazi. Ababhali balo mbono yilaba: uGeorge Siemens, owabeka imibono yakhe ephepheni elithi Connectivism: A Theory of Learning for the Digital Age, kanye noStephen Downes. Ikhono elibalulekile lapha yikhono lokusebenzisa kahle intuthuko yezobuchwepheshe kanye nokuthola ulwazi kuzinqolobane zangaphandle (lokho okubizwa ngokuthi ukwazi kuphi), hhayi olwazini olufundiwe ohlelweni lokufunda, kanye nekhono lokuzihlanganisa nokuzixhumanisa nolunye ulwazi.
Ezingeni le-neural, ithiyori ichaza amaqembu ama-neuron akha imihlangano eyinkimbinkimbi futhi exhunyiwe ebhekana nemiqondo eyisisekelo nolwazi. Ngokucwaninga ngezilwane ezihlolwayo ngama-electrode, ososayensi bathola ukuthi le "mihlangano" ye-neural ichazwe kusengaphambili ngezinhlobo ezithile zemisebenzi. Lokhu kudala uhlobo lwe-algorithm yobuchopho enokuxhumana okuthile okunengqondo. Ososayensi banethemba lokuthi ubuchopho bomuntu, nazo zonke izinkinga zabo, abusebenzi ngendlela ehlukile kunobuchopho bamagundane aselabhorethri.
Ubuchopho obuvela kuma-memristors
Uma sesiwazi kahle ama-algorithms, mhlawumbe ama-memristors angasetshenziswa ukulingisa ngokomzimba ubuchopho bomuntu. Ososayensi baseNyuvesi yaseSouthampton basanda kufakazela ukuthi bawusizo kulokhu.
Izikhumbuzo zososayensi baseBrithani, ezenziwe ngama-metal oxides, zasebenza njengama-synapse okwenziwa okufunda (kanye nokufunda kabusha) ngaphandle kokuphazamiseka kwangaphandle, kusetshenziswa amasethi edatha futhi aqukethe ulwazi oluningi olungabalulekile, njengoba kwenza abantu. Njengoba ama-memristors ekhumbula izifunda zawo zangaphambilini lapho ecishiwe, kufanele asebenzise amandla amancane kakhulu kunezinto ezijwayelekile zesekethe. Lokhu kubaluleke kakhulu ngokwenani lamadivayisi amancane angakwazi futhi akufanele abe nebhethri elikhulu.
Yiqiniso, lokhu kumane kuyisiqalo sokuthuthukiswa kwalobu buchwepheshe. Uma i-AI ibingalingisa ubuchopho bomuntu, ibingadinga okungenani amakhulu ezigidigidi zama-synapse. Isethi yama-memristors asetshenziswe abacwaningi yayilula kakhulu, ngakho-ke yayilinganiselwe ekufuneni amaphethini. Kodwa-ke, iqembu laseSouthampton liphawula ukuthi uma kunezicelo ezincane, angeke kudingeke ukuthi kusetshenziswe inani elikhulu kangaka lama-memristors. Ngenxa yabo, kuzokwazi ukwakha, isibonelo, izinzwa ezizohlukanisa izinto futhi zihlonze amaphethini ngaphandle kokungenelela komuntu. Imishini enjalo izoba usizo ikakhulukazi ezindaweni okunzima ukuzifinyelela noma eziyingozi kakhulu.
Uma sihlanganisa okutholwe okuvamile okwenziwe yi-Human Brain Project, imephu ye-"connectomes", ukuqashelwa kwe-intelligence algorithms kanye nobuchwepheshe be-memristor electronics, mhlawumbe emashumini eminyaka ezayo sizokwazi ukwakha ubuchopho bokwenziwa, ikhophi eliqondile. womuntu. Kwazi bani? Ngaphezu kwalokho, ikhophi yethu yokwenziwa cishe ilungiselelwe kangcono ukuguqulwa komshini kunathi.
