Jagoran Harshe AI

Tsarin Mulki AI

Tsarin tsarin mulkin AI shine hanyar Anthropic don daidaita samfura ta amfani da rubutattun ka'idoji - 'Tsarin Mulki' - don haka AI ta soki tare da sake duba nata amsoshin maimakon dogara ga mutane kawai don lakafta abun ciki mai cutarwa.

2 min karatuAn sabunta ta ƙarshe

Dubawa

It aims to make models helpful and harmless with far less human labor.

Zurfafa nutsewa

Daidaito na al'ada ya dogara kan ƙarfafa koyo daga ra'ayin ɗan adam (RLHF), inda mutane ke ba da fifikon samfura da yawa, gami da masu tayar da hankali, don koyar da ƙirar abin da za su guje wa. Kundin tsarin mulki AI yana rage wannan nauyi ta hanyar ba samfurin jerin ƙa'idodin rubuce-rubucen da aka zana daga tushe kamar sanarwar Majalisar Dinkin Duniya game da 'yancin ɗan adam da mafi kyawun ayyuka na aminci da aminci. Horon yana da matakai biyu. Na farko, matakin da ake kulawa: samfurin yana haifar da amsa, sannan ya soki shi a kan ka'idar tsarin mulki kuma ya sake rubuta shi don mafi kyau; Ana amfani da waɗannan amsoshin da suka inganta kansu don daidaita shi. Na biyu, matakin ƙarfafa-koyo, RLAIF, inda samfurin kansa ya ɗauki nau'i-nau'i na martani bisa ga tsarin mulki, kuma bayanan fifikon AI da aka ƙirƙira yana horar da samfurin lada. Ka'idodin a bayyane suke kuma ana iya daidaita su, suna mai da ƙimar jagorar ƙima ta zama abin dubawa maimakon ɓoyayye a cikin tambarin ɗan adam.

Fahimtar Fasaha

Ana kiran matakan biyu sau da yawa SL-CAI da RL-CAI. A cikin ilmantarwa da ake kulawa, madauki na 'zargi-da-bita' yana sa samfurin ya nemo inda amsarsa ta saba wa ƙa'idar da aka ƙirƙira ta sake rubuta ta, tana samar da bayanan horo ba tare da yiwa ɗan adam lakabin cutarwa ba. A cikin lokaci na RL, samfurin na biyu yana yin hukunci wanda daga cikin martani biyu mafi kyau ya bi tsarin mulki, yana samar da alamun fifikon AI (RLAIF) waɗanda ke horar da samfurin lada da aka yi amfani da su a daidaitaccen RL. Kundin tsarin mulki jagora ne bayyanannen rubutu wanda aka shigar cikin hanzari, don haka canza halayen ƙirar na iya zama kai tsaye kamar gyara ƙa'idodi.

Dabarun Tasiri

Gudu da sikelin

Gudun aikin harshe na iya tafiya da sauri ba tare da sadaukar da daidaito ba.

Shiga ku isa

Yana faɗaɗa damar shiga cikin harsuna da salon sadarwa.

Shawarwari masu haske

Ƙungiyoyi za su iya ciyar da ƙarin lokaci akan hukunci yayin da aiki da kai ke sarrafa maimaitawa.

Makomar Tsarin Mulki AI

Tsarin tsarin mulkin AI yana nuni zuwa ga 'sa ido mai daidaitawa,' inda AI ke taimakawa kula da AI yayin da samfuran ke girma da ƙarfi ga ɗan adam don bincika kowane fitarwa. Yi tsammanin mafi arziƙi, ƙarin ƙa'idodin tsarin mulki, shigar jama'a da shigar da jama'a waɗanda aka zaɓi ƙa'idodi a ciki (Anthropic sun gudanar da gwaje-gwajen 'Tsarin Tsarin Mulki na AI'), da hanyoyin haɗin gwiwar haɗa ra'ayoyin ɗan adam tare da zargi na AI. The transparency of written principles makes this attractive to regulators and auditors wanting to see the values a system encodes. Yayin da samfuran kan iyaka suka ci gaba, hanyoyin da ke barin samfuran su dogara da kansu da inganta kansu daga fayyace ƙa'idodi za su zama tsakiyar aminci.

Aiwatar da Gaskiyar Duniya

Horar da chatbot don ƙin taimakawa wajen kera makami ta hanyar yin sukar daftarin amsar nasa akan ƙa'idar gujewa cutarwa da sake rubuta shi.

Maye gurbin ɗan adam mai tsadar alamar jan ƙungiyar masu guba tare da bayanan fifikon AI da aka ƙirƙira (RLAIF) wanda tsarin mulki ke jagoranta.

Gyara ƙa'idar da aka rubuta don daidaita yadda tsattsauran ra'ayi ke da ƙima, sa'an nan kuma lura da canjin ɗabi'a ba tare da sanya dubban misalai ba.

Gudanar da ayyukan shigar da jama'a inda jama'a ke ba da shawarar ƙa'idodin da suka tsara tsarin tsarin tsarin

Hatsari & Tsare-tsare

Abubuwan da aka ruɗe suna iya shigar da rahotanni cikin nutsuwa, kwararar tallafi, ko abubuwan bincike.

Hankali na gaggawa na iya ƙirƙirar sakamako mara daidaituwa a cikin buƙatun iri ɗaya.

Za a iya fallasa bayanan rubutu mai ma'ana idan ikon samun dama yana da rauni.

Taswirar Hanya

1

Ƙayyade tsarin fitarwa, sautin, da ma'auni masu inganci kafin fitowa.

2

Amsa a ƙasa tare da amintattun tushe a duk lokacin da daidaito ya shafi mahimmanci.

3

Ajiye wurin binciken ɗan adam don abubuwan da ake samu masu girma.

4

Bibiyar tsarin gazawar kuma sake horar da tsokaci ko tafiyar aiki akai-akai.

Ci gaba da Bincike

Free newsletter

Get the daily AI briefing

Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.

One email each weekday. Unsubscribe in one click. We never sell or share your address.

Test yourself

Take the Constitutional AI quiz

Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.

Fara tambayoyi

Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation

Tambayoyin da ake yawan yi

What is Constitutional AI?

Tsarin tsarin mulkin AI shine hanyar Anthropic don daidaita samfura ta amfani da rubutattun ka'idoji - 'Tsarin Mulki' - don haka AI ta soki tare da sake duba nata amsoshin maimakon dogara ga mutane kawai don lakafta abun ciki mai cutarwa. Yana nufin yin samfura masu taimako da marasa lahani tare da ƙarancin aikin ɗan adam.

Menene 'Tsarin Mulki' a cikin Tsarin Mulki AI?

Kundin tsarin mulki tsari ne na rubutattun ka'idoji, wanda aka samo daga tushe kamar takaddun haƙƙin ɗan adam, waɗanda tsarin ke amfani da su don kimantawa da haɓaka amsoshinsa.

Wace babbar matsala tare da daidaitaccen RLHF Tsarin Tsarin Mulki AI yana son ragewa?

Ta hanyar samun samfurin ya soki kanta akan ƙa'idodi, Tsarin Tsarin Mulki AI ya rage ayyukan ɗan adam na yin bita da lakafta abubuwan da ke damun su ko cutarwa.

A cikin matakin kulawa na Tsarin Tsarin Mulki na AI, menene samfurin ke yi wa nasa fitarwa?

Samfurin yana gudanar da madauki-da-bita: yana gano inda daftarin sa ya saba wa ƙa'idar samfurin, sannan ya sake rubuta amsar, ƙirƙirar bayanan horo na inganta kansa.

Menene RLAIF ke tsayawa a matakin ƙarfafa-koyo?

RLAIF yana nufin Ƙarfafa Koyo daga AI Feedback: samfurin yana ba da amsa bisa ga tsarin mulki, yana samar da bayanan fifiko na AI maimakon alamun ɗan adam.

Me yasa ake ɗaukar amfani da ƙa'idodin da aka rubuta a matsayin fa'ida don bayyana gaskiya?

Saboda an rubuta ƙimar jagora, ana iya bincika su, a duba su, da kuma gyara su kai tsaye, ba kamar abubuwan da aka zaɓa ba a fakaice da ke cikin tarwatsa kimar ɗan adam.