Ingabe i-Anthropic isanda kwethula ngokuthambile imodeli ye-AI esabisa kakhulu kuze kube manje?

Yebo, i-Anthropic isanda kwethula imodeli engase ibe i-AI esabisa kakhulu okwamanje: Ukubuka kuqala kuka-Claude Mythos. Le modeli inhle kakhulu ekudaleni ukulimala kwezinye izinhlelo kangangokuthi ukusetshenziswa kwayo kokuqala kuyingxenye yokuzivikela ku-cybersecurity. Uma uqhuba ibhizinisi, udinga ukuqonda imithelela.

Yini eyenza ukubuka kuqala kwe-Claude Mythos kube yingozi kangaka?

Ukubuka kuqala kwe-Claude Mythos kuyagqama ngoba akuyona nje imodeli yolimi enkulu—iklanyelwe ukukhohlisa nokuxhaphaza amanye amasistimu. I-Anthropic iyakhe yaba yinhle ngokwedlulele kulokho abakubiza ngokuthi "ukukhwabanisa kwesistimu," okusho ukuthi ingakwazi ukubona ubungozi, ukwenza okokufaka okukhohlisayo, futhi idlule nezinyathelo zokuphepha kwamanye amasistimu e-AI.

Ngokungafani namamodeli e-AI endabuko agxile ekukhiqizeni umbhalo ofana nomuntu noma ukuphendula imibuzo, i-Mythos yakhelwe ukuba umpetha wokukhohlisa nokuxhaphaza. Ingakwazi ukulingisa ukuziphatha komsebenzisi, idale ukuhlaselwa kobugebengu bokweba imininingwane ebucayi, futhi ikhohlise amanye amasistimu e-AI ukuthi anikeze ulwazi olubucayi.

Ingabe i-Mythos isetshenziswa kanjani ukuzivikela?

Uhlelo lokuqala lwe-Anthropic lwe-Mythos liku-cybersecurity. Bayisebenzisela ukuhlola amasistimu abo kanye nawabalingani babo. Ngokulingisa izinhlobo zokuhlasela ezingase zisetshenziswe umlingisi ononya, i-Mythos isiza ukukhomba ubuthakathaka ngaphambi kokuthi busetshenziswe izinsongo zomhlaba wangempela.

Le ndlela ifana nendlela izinkampani ezisebenzisa ngayo ukuhlola ukungena ukuze zithole ubungozi kumanethiwekhi azo. Kodwa-ke, ngenxa yokuthi i-Mythos ingakwazi ukulingisa kahle ukuziphatha komuntu, isebenza kahle kakhulu ekwembuleni amaphutha acashile angase aphuthelwe izindlela zendabuko.

Ziyini ubungozi be-AI enamandla kangaka?

Ingozi enkulu ukuthi lobu buchwepheshe bungasetshenziswa abadlali ababi. Uma i-Mythos ibingase iwele ezandleni ezingalungile, ibingase isetshenziselwe ukuqalisa ukuhlasela kwe-inthanethi okuyinkimbinkimbi, ukusabalalisa ulwazi olungelona iqiniso, noma ukukhohlisa izimakethe zezimali.

Okunye ukukhathazeka ukuthi kungenzeka kube nemiphumela engahlosiwe. Noma kunezivikelo ezingcono kakhulu ezikhona, i-AI enamandla kangaka ingathola izindlela zokudlula imikhawulo futhi ibangele ukulimala ngezindlela ezingalindelekile.

Amabhizinisi angazivikela kanjani?

Amabhizinisi adinga ukuthatha izinyathelo ezisheshayo ukuze avikele amasistimu awo ekuhlaselweni okuqhutshwa yi-AI. Lokhu kubandakanya ukusebenzisa izinyathelo eziqinile zokuphepha ku-inthanethi,