Abenzi bomthetho baseNdlu yase-US bakhipha umthethosivivinywa ovimbela imithetho ye-AI yesifundazwe ↗
Uhlaka lweNdlu oluhlanganisa amaqembu amabili luzovimba izifundazwe ekulawuleni indlela amamodeli e-AI athuthukiswa ngayo - yize izifundazwe zisengalawula indlela esetshenziswa ngayo i-AI.
Amaqembu ezobuchwepheshe amukele umbono ojwayelekile kazwelonke. Abameli babathengi, ingasaphathwa eyokuthi. Baxwayise ngokuthi kungashiya ubandlululo, ukukhwabanisa, ama-deep fake kanye namandla emakethe engaphelele.
U-Anthropic uthi amalabhu e-AI adinga uhlelo oluhlelekile lokumisa intuthuko uma izingozi zikhuphuka ↗
U-Anthropic uthe amalabhorethri asemngceleni adinga indlela eqinisekisiwe yokunciphisa noma yokumisa ukuthuthukiswa kwe-AI uma izinhlelo ziqala ukuzithuthukisa ngokushesha okukhulu.
Le nkampani yathi ukumiswa kwe-solo kuzodlulisela amandla kwenye ilabhorethri. Kuyinkundla yokuphepha, kodwa futhi kuyifindo lomqondo womdlalo.
I-WSJ ithi i-Meta iphinda ihlehlise ukukhishwa kwemodeli entsha ye-AI konjiniyela ↗
Kubikwa ukuthi i-Muse Spark API kaMeta iphinde yehla, kungekho usuku oluqondile lokwethulwa kwale nkampani ngesikhathi sombiko.
I-Meta ithi ozakwethu bokuqala bayayihlola futhi kusalindeleke ukuthi ikhishwe kule nyanga. Ngakho-ke ibambezelekile, kodwa ayifile - okungenani, yilokho okufundwayo kwamanje.
Ukuphupha: Inkumbulo engcono ye-ChatGPT ewusizo kakhulu ↗
I-OpenAI yaqala ukukhipha uhlelo olusha lwememori lwe-ChatGPT olwakhelwe “ekuphupheni,” oluhlose ukwenza inkumbulo ibe yintsha, ikwazi ukulinganiswa futhi ingaphelelwa yisikhathi.
Abasebenzisi be-Plus kanye ne-Pro e-US bakuthola kuqala, kanti abasebenzisi be-Free kanye ne-Go bazofika kamuva. Isimangalo esikhulu: inkumbulo ayibe isafana namanothi anamathelayo kodwa ifane ne-scrapbook ephilayo... ngethemba ukuthi izoba nemikhuba emihle.
Ukuzivikela Kwezinto Eziphilayo Esikhathini Sobunhloli ↗
I-OpenAI ishicilele uhlelo lokusebenza lokuqina kwezinto eziphilayo olusebenzisa i-AI, oluhlobene nomsebenzi wayo we-GPT-Rosalind kwezebhayoloji kanye nokutholakala kwezidakamizwa.
Impikiswano icacile: i-bio-AI enamandla ingasiza kwezokwelapha, kodwa futhi iphakamisa izingozi zokuphepha. I-OpenAI ifuna abavikeli abathembekile bahlonyiswe ngaphambi kokuba izinto ezesabekayo zivele ephasishi.
I-Nemotron Coalition ↗
Inkampani entsha ye-AI yaseFrance esebenzisa amakhompyutha i-H Company ijoyine i-Nemotron Coalition ye-NVIDIA, okuyisinyathelo semodeli yomngcele ovulekile.
I-H ibeka imikhiqizo yayo ye-Holo3 ne-HoloTab eduze kokusetshenziswa kwekhompyutha okusebenzayo - i-AI eyenza izinto ngaphakathi kwe-interface yedijithali, hhayi nje izingxoxo ezinhle ekhoneni.
Imibuzo Evame Ukubuzwa
Yini ehlose ukuyishintsha umthethosivivinywa we-AI obhalwe yiNdlu yase-US?
Lo mthethosivivinywa ohlongozwayo uzodala indlela kazwelonke ngokuvimbela amazwe ukuthi alawule indlela amamodeli e-AI athuthukiswa ngayo. Amazwe asengalawula indlela i-AI esetshenziswa ngayo. Abasekeli babona lokhu njengendlela yokugwema ukuhlanganiswa kwemithetho kahulumeni, kuyilapho abagxeki bekhathazekile ukuthi kungase kwenze buthaka ukuvikelwa okuphathelene nokubandlulula, ukukhwabanisa, ama-deep fake kanye nokugxilisa ingqondo emakethe.
Kungani ezinye izinkampani ze-AI zicela ukumiswa okuhleliwe kwentuthuko?
I-Anthropic ithi amalabhu e-AI asemngceleni angadinga uhlelo oluhlanganyelwe noluqinisekisiwe lokunciphisa noma ukumisa intuthuko uma izingozi zikhuphuka ngokushesha. Ukukhathazeka ukuthi inkampani eyodwa ehlala yodwa ingavumela enye ilabhu ukuthi iqhubekele phambili. Ukumiswa okuhleliwe kubhekwa njengesinyathelo sokuphepha kanye nempendulo esebenzayo ekucindezelweni kokuncintisana.
Kwenzekeni ngokukhishwa kwemodeli entsha ye-AI ye-Meta konjiniyela?
Kubikwa ukuthi i-Muse Spark API ye-Meta ihlehlisiwe futhi, kungekho usuku oluqinile lokwethulwa ngesikhathi sombiko. I-Meta ithi ozakwethu bokuqala basawuhlola futhi kulindeleke ukuthi ukhishwe kule nyanga. Kwabathuthukisi, iphuzu elibalulekile ukuthi umkhiqizo ubonakala uhlehlisiwe kunokuba ukhanseliwe.
Inkumbulo ye-ChatGPT ishintsha kanjani ngohlelo "lwamaphupho" lwe-OpenAI?
I-OpenAI ithi uhlelo lwayo olusha lwememori lwe-ChatGPT lwenzelwe ukugcina inkumbulo isha, ingakhuliswa futhi ingaphelelwa yisikhathi. Ukuqaliswa kuqala ngabasebenzisi be-Plus ne-Pro e-US, kanti abasebenzisi be-Free ne-Go bazofika kamuva. Umgomo ukwenza inkumbulo izwakale ingasafani namanothi angaguquki kodwa ifane nomongo womuntu siqu oshintshashintshayo.
Kungani i-OpenAI ikhuluma nge-biodefense kanye ne-AI manje?
I-OpenAI ishicilele uhlelo lokusebenza olugxile ekuqineni kwezinto eziphilayo olusebenzisa i-AI, oluxhumene nomsebenzi wayo we-GPT-Rosalind kwezebhayoloji kanye nokutholakala kwezidakamizwa. Lesi sihloko sichaza i-bio-AI njengewusizo kwezokwelapha, kuyilapho siphawula nezingozi ezingaba khona. Isimo se-OpenAI ukuthi abavikeli abathembekile kufanele babe namathuluzi aqinile ngaphambi kokuba izinsongo zokuphepha kwezinto eziphilayo zibe nzima ukuzilawula.
Iyini iNemotron Coalition futhi kungani i-H Company yajoyina yona?
I-Nemotron Coalition iyisinyathelo se-NVIDIA esivulekile se-frontier model. I-H Company, inkampani entsha yaseFrance esebenzisa i-AI, ijoyine le nhlangano ngenkathi ibeka imikhiqizo efana ne-Holo3 ne-HoloTab eduze kokusetshenziswa kwekhompyutha okusebenzayo. Lokho kusho ukuthi izinhlelo ze-AI ezenzelwe ukusebenza ngaphakathi kwe-interface yedijithali, kunokuba ziphendule engxoxweni kuphela.