Ngokombiko we-TechCrunch, ngesikhathi sendawo ngomhlaka-20, i-Pew Research Center ikhiphe ucwaningo olukhombisa ukuthi ngaphezu kwengxenye yesithathu yamakhasi ewebhu ashicilelwe kusukela kwavela i-ChatGPT ngoNovemba 2022 kungenzeka anezimpawu zokwenziwa yi-AI. Lo mphumela uqinisekiswa olunye ucwaningo, ukhombisa ukuthi inani elikhulu lamakhasi ewebhu engezwe eminyakeni yakamuva selakhiwe yi-AI ngokuqondile noma “lilungiswe kakhulu” yi-AI.
Ngaphambi nje kokushicilelwa kwalolu cwaningo, umhlinzeki wezinsizakalo zengqalasizinda ye-inthanethi i-Cloudflare waveza ukuthi ithrafikhi yewebhu ekhiqizwa amarobhothi isidlule ithrafikhi yabantu, futhi lokhu kwenzeke ngaphambi kunalokho obekubikezelwe.
Ucwaningo lwe-Pew Research Center lugxile embuzweni wokuthi “yikuphi okuqukethwe okutholakala ku-inthanethi okufinyelelwayo”. Imiphumela ikhombisa ukuthi amakhasi amaningi ewebhu kungenzeka akha umjikelezo lapho amarobhothi efunda okuqukethwe okwenziwe amarobhothi.
Ukuze yenze lolu cwaningo, i-Pew Research Center yasebenzisa ingobo yomlando yamakhasi ewebhu ye-Common Crawl yaqoqa cishe amakhasi ewebhu esiNgisi angu-500,000 eminyakeni engaba ngu-5 edlule. Abacwaningi base besebenzisa ubuchwepheshe be-Open Pangram ukuhlola ukuthi la makhasi kungenzeka yini abhalwe yi-AI noma aguqulwe kakhulu yi-AI.
Kwisampula yamakhasi ewebhu angu-10,000 akhethwe ngokungahleliwe ngoJulayi 2026, cishe u-10% wamakhasi ukhombise izimpawu ezicacile zokudalwa yi-AI.
I-Pew Research Center iveze ukuthi isampula ekhethwe ngokungahleliwe nakanjani izofaka amanye amakhasi ashicilelwa kudala, ngesikhathi amathuluzi okubhala e-AI ayengakabi khona, ngakho la makhasi ayengenakwenziwa yi-AI.
Ngakho-ke, ukuze ihlole ngokunembe kakhulu umthelela we-AI kokuqukethwe kwe-inthanethi, i-Pew Research Center yasusa amakhasi amadala ashicilelwa ngaphambi kokukhishwa kwe-ChatGPT, yahlaziya kuphela okuqukethwe okusha okwengezwe ngemva kokuvela kwe-ChatGPT.
Kwisampula elungisiwe, u-35% wamakhasi ewebhu ubonise izimpawu zokudalwa yi-AI. Ngokohlobo lwesizinda, isilinganiso samasayithi e-.com abonisa izimpawu zokudalwa yi-AI sicishe sibe izikhathi ezingu-10 kunesezingosi ze-.edu ne-.gov. Kulezi zinhlobo ezimbili zokugcina zamasayithi, okuqukethwe okwenziwe yi-AI kucishe kube ngu-1%. Isilinganiso samasayithi e-.org sona singu-4.6%.
I-Pew Research Center iyavuma ukuthi lolu hlaziyo lunemikhawulo. Njengamanye amathuluzi okuhlola i-AI, i-Pangram ingase yenze amaphutha, ibone amakhasi abhalwe abantu njengokungathi akhiqizwe yi-AI. Nokho, uma kubhekwa isampula enkulu, imiphumela yocwaningo isakwazi ukukhombisa umkhuba jikelele.
Ucwaningo luphinde lwathola ukuthi ezinye izindlela zokubhala ezivame ukubhekwa njengezimpawu zokubhala nge-AI sezivame kakhulu eminyakeni yakamuva: amadeshi amade, ikhoma yase-Oxford, kanye nemisho efana nokuthi “lokhu akuyona i-X, kodwa yi-Y”.
Qaphela: Ikhoma yase-Oxford (Oxford Comma), ebizwa nangokuthi ikhoma elandelanayo (Serial Comma), yikhoma esetshenziswa ekubhaleni kwesiNgisi phakathi kwento engaphambi kowokugcina nesihlanganiso esifana no-and noma u-or ohlwini oluqukethe izinto ezintathu noma ngaphezulu. Isibonelo, ku-“A, B, and C”, ikhoma elandela u-B yikhoma yase-Oxford.