Sekungamashumi eminyaka, i-IBM ingumholi ongenambangi wemboni yekhompyutha. Ngisho nanamuhla, nakuba ingayithathi indima enkulu eyake yayenza, umsebenzi wakhe usaqhubeka nokusebenza. Kodwa-ke, esimweni se-Artificial Intelligence, ukungena nokuphuma kwe-IBM kwakushesha kakhulu nakuba kubaluleke kakhulu.
Eminyakeni yamashumi amahlanu, ucwaningo kulo mkhakha lwaluthathwa njengengxenye ebalulekile yolwazi oludingekayo ukuze kwakhiwe amakhompyutha angcono. Ngakho i-IBM yanquma ukunikeza ukukhanya okuluhlaza ekuthuthukisweni kohlelo oluzoxazulula ama-theorems.
Ukungena nokuphuma kwe-IBM
sabona ku izindatshana ezedlule ukuthi uSimon nethimba lakhe baphumelele ngohlelo olwafakazela amathiyori ezibalo. Ngenxa yalokhu, bekufanele bakhe ulimi olusha lokuhlela.
Inselele ebhekene ne-IBM ibe nenkinga eyengeziwe. Ukuze ikhompuyutha ifakazele i-theorem yejometri, kwakudingeka ibone isibalo. Amakhamera ewebhu nesofthiwe ekwazi ukucubungula izithombe bekungakaqanjwa.
Ithimba elikhethelwe umsebenzi Bekuyi-IBM 704. Ithathwa njengomunye wamakhompiyutha amakhulu okuqala emhlabeni fBekungokokuqala ukufaka indawo yokuntanta yehadiwe.
Imisebenzi yephoyinti elintantayo ihlanganisa ukuhlanganisa, ukususa, ukuhlukanisa, ukuphindaphinda, kanye nezibalo zempande eyisikwele ezinezinombolo ezinkulu kakhulu noma ezincane kakhulu.
Iphinde ibe nenkumbulo eyinhloko kazibuthe eyayishesha kunesistimu yezigubhu kazibuthe esetshenziswe kumamodeli wangaphambilini futhi ikwazi ukwenza imiyalelo eyizinkulungwane ezingamashumi amane ngomzuzwana evezwe emiyalweni engu-36-bit.
Isofthiwe yathatha iminyaka emithathu ukuba ilungele njengomphathi wayo, udokotela we-physics okuthiwa u-Herbert Gellenter, tKwadingeka asungule ulimi olusha lokuhlela engashintsha izimpawu ezifana ne-IPL kodwa ebinohlelo olulula lwe-Fortran, intuthuko eyenziwe yi-IBM ngokwayo yokwakhiwa kwezicelo zokubala kwesayensi.
Uhlelo ithole ulwazi mayelana nesibalo sejiyomethri okwakufanele sisebenze ngayo ngendlela yochungechunge lwezixhumanisi ezifakwe ngendlela yamakhadi ashaywayo futhi wayekhipha imiphumela emaphakathi kudatha eyaziwayo.
Isofthiwe yaziwa ngokuthi i-Geometry Theorem Prover (I-Geometry Problem Prover) futhi isekelwe kulezo zakhiwo eyayingaziqinisekisa emdwebeni. wakwazi ukuxazulula ithiyori enezinyathelo ezimbili kuya kumathuba ahlukene angamashumi amabili nanhlanu lapho, ekwenza ngokungaboni, kuzodingeka ahlaziye isigidi.
Umhloli wenkinga yejiyomethri ube ngowokuqala ukusebenzisa indlela eyaziwa ngokuthi imodeli yereferensi.. Ngaphandle uma ufika esosweni elindizayo emizuzwini engu-5 edlule, ngokuqinisekile uzwile ngomunye wemiphumela yakamuva yokusetshenziswa kwalolu hlelo lokusebenza: ChatGPT.
Emkhakheni wobuhlakani bokwenziwa Imodeli iwumfanekiso wenqubo eyinkimbinkimbi lapho okucatshangwayo kungenziwa khona. Esimeni somhloli, imodeli bekuyizixhumanisi zesibalo sejiyomethri, esimweni se-ChatGPT imodeli ekwazi ukucubungula ulimi lwabantu.
Amanye amakhompyutha e-IBM ayenza izinto ezingathí sina njengokufunda okuhlola noma i-chess. Esimweni sokuqala, inhloso kwakuwukubona ukuthi umshini unekhono yini lokufunda indlela odlala ngayo. Ugcine ekwazile ukumshaya ngakho kubukeka sengathi wenzile.
Ukuphumelela kwasekuqaleni kwe-IBM nakho kwaba yimbangela yokushiya kwayo lo mkhakha. Amakhompiyutha anikezelwe ukudlala i-chess namasheke aphumelele kakhulu ngomshini wokunyathelisa, kodwa hhayi kakhulu phakathi kwabanikazi bamasheya benkampani ababekubheka njengokumosha imali.
Kulokhu kumele sengeze ukuthi umnyango wokumaketha wenkampani Waqaphela ukuthi phakathi kwamakhasimende akhe ayengase abe nokungathembi amakhompyutha kwakukhula. Bekunovalo kulaba ababhekelele ukuwathatha ngokuthi ukuthengwa kwawo kuzogcina kuthathe indawo yawo.
Ucwaningo ku-Artificial Intelligence yayekwa futhi isu elisha lokumaketha kwakuwukuphawula amakhompyutha njengabacubunguli bedatha abenza lokho abakutshelwayo kuphela.
Ingabe okufanayo kuzokwenzeka ngamathuluzi amasha manje ethandwa kakhulu? Ngabe bazovalwa ezinkampanini ngoba besaba ukuthi bazogcina bengene ezikhundleni eziphezulu?
Kuzomele silinde ukuze sikubone.