Ngemva kokwenza izinsiza zememori nezokugcina idatha zibe ngaphansi kwengcindezi enkulu, izinkampani ze-AI manje kubonakala sengathi sezibhekise amehlo azo efeni lezincwadi zesintu. Imithombo yezindaba ephenyayo i-404 Media ibike muva nje ukuthi izinkampani eziningana ze-AI zithenga ngobuningi izincwadi ezisetshenzisiwe ngabalamuli, ukuze zithole idatha yokuqeqesha esezingeni eliphezulu yokuqeqesha amamodeli e-AI, kuyilapho zigwema ukuvusa ukusabela okubi komphakathi.
Ukuthuthuka kwe-AI kuncike enanini elikhulu ledatha, kodwa akuyona yonke idatha enenani; okubalulekile yikhwalithi yedatha. Eminyakeni yamuva, okuqukethwe okuningi okusezingeni eliphansi okukhiqizwe yi-AI, okwaziwa ngokuthi “okuqukethwe okungenamsebenzi kwe-AI”, kugcwele ku-inthanethi. Lokhu akungcolisi nje kuphela indawo yedatha, kodwa futhi kwehlisa ukusebenza kwe-AI ekuqhubekeni nokufunda. Ngenxa yalokho, izinkampani ezihamba phambili ze-AI seziqale ukufuna kabusha okuqukethwe okudalwe ngabantu, ikakhulukazi izincwadi eziphrintiwe ezashicilelwa ngaphambi kuka-2022, ngoba kunamathuba amakhulu okuthi lezi ziqu zibe yimisebenzi yokuqala futhi zingakathintwa okuqukethwe okukhiqizwe yi-AI.
Empeleni, izinkampani ze-AI sezike zasebenzisa izincwadi eziphrintiwe ukuqeqesha amamodeli. U-Anthropic, ongomunye wezinkampani ezinkulu ze-AI ezibandakanyeka manje emacaleni okuphulwa kwamalungelo obunikazi, wake wasebenzisa izigidi zama-dollar ukuthenga inqwaba yezincwadi eziphrintiwe ukuze akhiphe okuqukethwe, aqeqeshe ngakho imodeli ye-Claude AI, abese ezicekela phansi lezo zincwadi. Kubikwa ukuthi le nkampani yathenga izincwadi ngobuningi kwa-Better World Books. Nakuba inkantolo yanquma ukuthi ukusebenzisa izincwadi ezisemthethweni ukuqeqesha i-AI kungena ngaphansi “kokusetshenziswa okufanele” komthetho wamalungelo obunikazi, u-Anthropic waphinde waphula amalungelo ababhali nabashicileli ngokugcina isikhathi eside isizindalwazi esasinezincwadi eziyizigidi ezingu-7 ezithathwe ngokungemthetho, wagcina egwetshwe ukukhokha isinxephezelo esifika ku-1.5 billion dollars (inothi: ngezinga lokushintshisana lamanje cishe elilingana nama-yuan ayizigidi eziyizinkulungwane ezingu-10.162).
Impikiswano efanayo yenzeke naku-Google. Muva nje, umfelandawonye wabashicileli uvulele i-Google icala, usola le nkampani ngokusebenzisa ngaphandle kwemvume izigidi zezincwadi ezivikelwe ngamalungelo obunikazi ukuqeqesha imodeli ye-Gemini AI.
I-ISBNdb, isizindalwazi esiku-inthanethi esinolwazi lwamakhathalogi ezincwadi ezingaphezu kwezigidi ezingu-111, sekuyisikhathi eside iyinkundla ebalulekile yabathengisi bezincwadi, imitapo yolwazi nabasabalalisi. Njengoba imboni ye-AI ikhula ngokushesha, i-ISBNdb nayo isiqalile ukulungisa indlela yebhizinisi layo, yethula isevisi yokuthenga izincwadi ngobuningi eqondiswe ezinkampanini ze-AI.
I-404 Media ibike ukuthi ubukhulu bama-oda busukela ekuthengweni kwezincwadi ezingu-1000 ngesikhathi esisodwa kuye ezincwadini eziyisigidi ngesikhathi esisodwa.
Umthengisi wezincwadi oqeqeshiwe ongathandanga ukudalulwa kwegama uthe kusukela ngo-Ephreli kulo nyaka, ukuthengiswa kwezincwadi kukhule ngendlela engakaze ibonwe. Esikhathini esidlule, lapho ibhizinisi lihamba kahle, wayekwazi ukuthengisa izincwadi ezingaba ngu-20 kuphela ngesonto; kodwa ezinyangeni ezimbalwa ezedlule, ukuthengiswa kwamasonto onke kukhuphuke kwaba ngamakhulu, okungaphezu kokuphindwe kahlanu kokujwayelekile.
Abanye abathengisi bezincwadi abasezinkundleni zokuhweba izincwadi ezisetshenzisiwe njenge-Alibris ne-Biblio nabo babike izigameko ezifanayo zokuthengwa ngobuningi.
Yize okwamanje bungekho ubufakazi obuqinisekile bokuthi lezi zenzo zokuthenga zivela ku-ISBNdb noma kwenye inkampani ethile ye-AI, kunezinto eziningi ezingajwayelekile. Ngokwesibonelo, cishe zonke izincwadi ezithengwa ngobuningi zine-International Standard Book Number (ISBN), ikhodi enamadijithi angu-13 esebenza njengophawu oluyingqayizivele lokuhlonza izincwadi emhlabeni wonke.
Ngaphezu kwalokho, lezi zenzo zokuthenga cishe azikhethi nhlobo isihloko, uhlobo noma umbhali. Izincwadi ezinjengamanoveli, izincwadi zokufunda nezincwadi zobungcweti zonke ziyathengwa. Okumangaza nakakhulu ukuthi abathengi kubonakala sengathi abayikhathaleli nhlobo intengo; ngisho nalapho ezinye izincwadi zibiza kakhulu kunentengo yemakethe, bayazithenga ngokuqondile.
Ngesikhathi kuqhubeka icala lamalungelo obunikazi lika-Anthropic, u-Tom Harvey, owake wabamba iqhaza kuphrojekthi ye-Google Books futhi kamuva waphatha iphrojekthi yokwenza amadokhumenti abe yidijithali ka-Anthropic, “Project Panama”, waqinisekisa ukuthi u-Anthropic wake waqasha izinkampani eziningana ezikhethekile ekuskeneni amadokhumenti ukuze zenze izincwadi eziphrintiwe zibe yidijithali.
Enye yezinkampani ezabambisana nayo yi-Datamation Information Services, ehlinzeka ngezinsizakalo ezimbili zokwenza izincwadi zibe yidijithali: ukuskena okungonakalisiyo nokuskena okonakalisayo.
Ukuskena okungonakalisiyo kuvame ukusebenzisa izikena ezibheka phansi, izikena eziyipulatifomu noma izikena ezimise okwe-V, ukuze izincwadi zenziwe zibe yidijithali ngaphandle kokuzihlakaza. Kanti ukuskena okonakalisayo kudiliza incwadi ngokuqondile, bese ikhasi ngalinye lifakwa kusikena sezimboni esinesivinini esikhulu ukuze licutshungulwe.
Ngenxa yokuthi ukuskena okonakalisayo kusebenza ngokushesha futhi kubiza kancane, izinkampani ze-AI zivame ukukhetha le ndlela, okuholela ekubhujisweni kwezigidi zezincwadi eziphrintiwe.
Akungabazeki ukuthi izincwadi eziphrintiwe ziyinqolobane enkulu yolwazi ku-AI. Nokho, le ndlela nayo ivuse izimpikiswano zokuziphatha eziqhubeka zanda.
Abagxeki bathi akukacaci ukuthi izinkampani ze-AI, ngesikhathi zenza izincwadi zibe yidijithali, ziyahlukanisa yini phakathi kwezincwadi ezijwayelekile, eziyigugu nezincwadi ezingasashicilelwa. Uma lezi zincwadi eziyivelakancane zibhujiswa, zingase zingaphinde zisatshalaliswe unomphela.
Enye inkinga enkulu ukuthi okuqukethwe okuskeniwe ekugcineni kuzongena ezizindalwazini ezizimele zezinkampani ze-AI, kusetshenziselwe ukuqeqesha i-AI kuphela, kuyilapho umphakathi ojwayelekile ungeke ukwazi ukukufinyelela.
Lokhu kusho ukuthi abantu bangase bathole i-AI ehlakaniphe kakhulu, kodwa intengo yalokho kungaba ukuthi inani elikhulu lezinsiza zolwazi eziqoqwe isintu ngeke lisashiyelwa izizukulwane ezizayo ngendlela evulekile nelufinyelelekayo.