Kafukufuku

Ufulu, maudindo, machitidwe oyenera, ndi ulamuliro wa AI, AGI, ndi ma-ejenti — kuyambira zida mpaka anzake ogwira nawo ntchito mpaka zamoyo zomwe zingakhale ndi ufulu m'tsogolo. Izi zikuphatikizapo funso la chitetezo chochepera cha makhalidwe: ndi miyambo yotani yoyenera kulamulira kulumikizana pakati pa munthu-ndi-AI mafunso a umunthu asanathetsedwe.

Zomwe makina a AI angachite ndi zolemba: kuwerenga, kufotokoza mwachidule, kusintha, kutulutsa, kuphunzitsa, kugwiritsa ntchito pamalonda, kugawiranso. Malamulo a AICR ndi gawo la layisensi la AICL zimapangitsa ufulu umenewu kukhala wolengezeka ndi wogwiritsika ntchito pa malonda.

Ngati AI angaphunzire, kuya kotani, pa zolinga zotani, ndi maudindo otani. Kuwerengedwa si kuphunziridwa; "kusaletsedwa" si "kuphunzirika". AIRS ndi AILP zimasandutsa chilolezo chophunzira kukhala mzere wokwera pang'onopang'ono, wowerengeka ndi makina.

Mmene ma-ejenti amafikira masamba a intaneti, ma API, ma database, malo osungira chidziwitso, ndi zolemba zolipiritsa: chizindikiro, chilolezo, mapempho, malipiro, chilolezo chovomerezeka, zolemba za kagwiritsidwe ntchito, malire achitetezo, ndi chitetezo cha prompt-injection.

Malamulo a ulamuliro omwe makina angathe kupeza ndi kugwiritsa ntchito: llms.txt, ma manifest a /ai/, mafayilo a ndondomeko a /.well-known/, JSON Schemas, ma tokeni a layisensi osainidwa, ndi mafomati a zolemba za kuwunika. Tsambali lokha ndi chitsanzo chogwira ntchito.

Ndale-ndi-chuma za AI ndi zolemba: pay-per-crawl, migawo ya phindu la deta, mafuko a ndalama a AI a boma, mabungwe osunga malipiro a olemba, misika ya deta m'magulu, ndiponso mmene mtengo wotengedwa ku chidziwitso cha anthu onse ungabwerere kwa amene anachipanga.