I-CEO ye-Anthropic ichaza uhlelo lokunciphisa intuthuko ye-AI
UDario Amodei usanda kucela imboni ukuthi yehlise ijubane. Ingami. Ijubane. Indatshana ende, izinyathelo ezintathu, kanye nesexwayiso sokuthi ukuzithuthukisa kabusha sekupheka ngokushesha kunalokho abantu abadala abangakuhlola.
I-Anthropic imema ngokuzithandela abahloli abahlanganisiwe ngaphakathi kwesakhiwo - amabheji, amadeski, ama-laptop, kanye nelungelo lokushicilela okutholakele ngaphandle kwesihlungi se-PR. Bese ifuna amalebhu entando yeningi ahlanganise amazinga okuphepha. Bese, ngandlela thile, isivumelwano nohulumeni abanamandla. Kulula ephepheni. Kunzima yonke indawo.
Usasho ukuthi i-AI ingaba isimangaliso. Uthi futhi iqembu elinamandla kakhulu elingahlelekile lingase lithathe i-inthanethi nge-botnet eqhubekayo futhi lenze umonakalo omkhulu wezigidigidi. Leyo misho emibili iseduze komunye nomunye futhi ayizwani.
O - futhi ufuna ukuxolelwa okuncane kokungathembeki ukuze izingxoxo zokuphepha zingabi yinkinga yezomthetho. Amabhuleki ahambisana namaphepha. Yebo ayahambisana nawo.
USam Altman uqinisekisa ukuthi i-OpenAI ngeke iphume emphakathini kulo nyaka, ethi i-IPO manje izofika 'ngesikhathi esingafanele' ngenxa yokukhathazeka ngokuphepha kwe-AI
U-Altman wahlala phansi ukuze enze inhlolokhono ekhethekile waqeda ikhalenda lokufakwa ohlwini. Uthe i-IPO njengamanje izoba "yisikhathi esingafanele." Umsebenzi wokuphepha awukaqedwa. Ukuqondanisa akukaqedwa. Umphakathi awukakulungeli. Ngesikhathi bemcindezela - hhayi kulo nyaka.
Amabhange asevele ejikeleza. Ukufakwa okuyimfihlo kukhona. Manje i-CEO ithi banezinto eziningi okufanele bazenze kuqala, okuhlanganisa nokuthola ukuthi imboni nohulumeni basebenzisane kanjani. Uzwakale sengathi ukhululekile ukulinda.
Uphinde wathi ukwakha i-AI ngaphezu kokulawulwa ngumuntu kungenzeka "ngokuphelele" - wabe esefunga ukuthi uzoyeka ukuqeqeshwa uma leyo yindlela. Okunesibindi. Futhi kulula, mhlawumbe. Mhlawumbe kokubili. Ukuhlangana kwenhlangano engenzi nzuzo/okwenza inzuzo, waphikisana ngokuthi, kukhona ukuze bakwazi ukwenza izingcingo ezingabalulekile kubanikazi bamasheya. Sizobona ukuthi leso sigwebo siyasinda yini ekuthinteni i-prospectus yangempela.
USam Altman we-OpenAI ukhomba isivumelwano nezinye izinkampani ze-AI ukubhekana nezingozi zokuphepha
Ingxoxo efanayo, isihloko esihlukile, futhi lena iwukuhleba. Ukuhlala phansi no-Amodei, uMusk, noDemis Hassabis kuyisinyathelo esilandelayo esingokwemvelo. U-Altman: ucabanga ukuthi lokho kuzokwenzeka. Ngeke amemezele izingxoxo zangasese kusengaphambili. Wabe esezimemezela kusengaphambili.
Ingozi enkulu enamadijithi amabili ayamukeleki, kusho yena. Amamodeli angakakhishwa asevele enekhono kangangokuthi ukuqhubekela phambili ngaphandle kokuqapha okungcono kuzwakala kungenangqondo. Isivumelwano sangempela kanye nesimo esivuvukele ngokucophelela kungabonakala kufana ngaphandle.
Abasebenzi base bezwile kakade inguqulo yalokhu ngaphakathi. Emphakathini kuseyinkungu. Uma igumbi livela, kuhle. Uma lingenzeki, ikhotheshini isasebenze kahle.
USam Altman uyawusekela umkhankaso kaDario Amodei wokunciphisa ijubane, futhi uthi i-OpenAI izokwenza okufanayo
Ngemva kwamahora amaningi ebhala indatshana, u-Altman wathumela ukuthi i-OpenAI izohambisana nesithembiso somhloli ohlanganisiwe. Ukufinyelela okufana nabasebenzi. Okuningi ongakwabelana ngakho maduze. UMusk wabhala amagama amathathu: "UDario uqinisile." Abaphathi abathathu abaphikisanayo. Isiqubulo esisodwa. Amandla empelasonto angathandeki.
Ukuxhumanisa kanye nemidlalo yeshashalazi kungahamba ndawonye. Isigaba sokuqala sohlelo luka-Amodei manje sinezindawo ezimbili ezinkulu zokucwaninga ezivuma ngekhanda esidlangalaleni. Ubuso Obugobile - yebo, lobo Buso Obugobile - bucele ukujoyina uhlelo lokuhlola futhi bethula into ebizwa ngokuthi i-Open Alignment Initiative. Ukucaca okuvela kubantu abagqekeziwe. Okunezimbongi, noma okukhonjwe.
IYurophu, noma ngabe ikufanele kangakanani, ayinamuntu ongacela lowo myalelo wokungakhokhiswa intela njengowase-US. IBrussels yayeka kudala ukunikeza abantu inkululeko yokungakhokhiswa intela ngabanye. Ngakho ukuxhawulana kungokweMelika. Okunye okusele kumephu… ukuhlikihla amahlombe.
I-OpenAI ifuna nje ukuwina
Izazi zezibalo azikho esimweni sokugubha ngemva kwempelasonto yezingxoxo. I-Navier-Stokes ye-OpenAI egijima - cishe ama-ejenti ayizinkulungwane eziyishumi, amashumi ezigidi zamaRandi e-computing, amahora angamashumi ayisishiyagalombili nesishiyagalombili - ibukeka njengokufuna izindebe ngaphakathi enkundleni. Izazi zezibalo zifuna ukuqonda. Ilebhu, baqhubeka bethi, ifuna ukuwina.
UTristan Buckmaster ubatshele ukuthi i-OpenAI inikeza ukubala okungenamkhawulo kanye nokubhala yedwa uma ingayishiya inhlangano yayo esebenzisana ne-Anthropic. Wayibiza ngokuthi ukufumbathisa. Wathi cha. I-OpenAI ithi, ngokweqiniso, izikhuthazo zayo ze-Codex bezingeke zithinte uhlelo. Akaqiniseki. U-Andreas Thom usazibuza ukuthi ngabe izingxoxo ze-ChatGPT ngomsebenzi wakhe zanikeza imodeli eyakha phezu kwayo kamuva. Yinkampani kuphela ezokwazi. Usola ukuthi abazi ngisho nokuthi.
Bese kuba yinkinga enkulu: I-OpenAI ithi isivele yenze intuthuko enkulu kwenye inkinga yeMillennium Prize. Engashiwongo igama. UClay usadinga iwindi elide lokwamukelwa komphakathi noma kunjalo. Ngakho-ke akekho "owinile" okwamanje. Bavele bagijimela olandelayo. Kuhle.
Imibuzo Evame Ukubuzwa
Iyini icebo lika-CEO we-Anthropic uDario Amodei lokusheshisa ukuthuthukiswa kwe-AI?
U-Amodei ushicilele indatshana ende ecela imboni ukuthi yehlise ijubane - hhayi ime - ngoba ukuzithuthukisa okuziphindaphindayo kuthuthuka ngokushesha kunalokho abahloli bezimali abangakulandela. Isinyathelo sokuqala simema abahloli abahlanganisiwe ku-Anthropic ngamabheji, amadeski, ama-laptop, kanye nelungelo lokushicilela okutholakele ngaphandle kwesihlungi se-PR. Ube esefuna amalabhorethri entando yeningi ahlanganise amazinga okuphepha, futhi ekugcineni, uhlobo oluthile lokusebenzisana nohulumeni abanamandla. Ufuna futhi ukuxolelwa okuncane kokungathembeki ukuze izingxoxo zokuphepha zingabi yinkinga yezomthetho.
Kungani uSam Altman athi i-OpenAI ngeke isebenzise i-IPO kulo nyaka?
Engxoxweni ekhethekile, u-Altman uthe i-IPO njengamanje izoba yisikhathi esingafanele ngoba umsebenzi wokuphepha, ukuvumelanisa, kanye nokulungela umphakathi akukapheli. Lapho ecindezelwa, uqinisekisile ukuthi ngeke kube njalo kulo nyaka yize amabhange ejikeleza futhi kukhona ifayela eliyimfihlo. Uphikise ngokuthi isakhiwo esingenzi nzuzo/senzuzo sikhona ukuze i-OpenAI ikwazi ukwenza izingcingo ezingabalulekile kubanikazi bamasheya. Uphinde wathi ukwakha i-AI ngaphandle kokulawulwa ngumuntu kungenzeka ngokuphelele futhi wafunga ukuthi uzoyeka ukuqeqeshwa uma leyo yindlela.
Ingabe i-Altman ibonise isivumelwano sokuphepha namanye amalabhorethri e-AI?
Kuleyo ngxoxo efanayo, u-Altman uthe ucabanga ukuthi ukuhlala phansi nontanga abanjengo-Amodei, uMusk, noDemis Hassabis kuzokwenzeka, kuyilapho enqaba ukumemezela izingxoxo zangasese ngaphambi kokumemezela. Uphikise ngokuthi ingozi enkulu kakhulu ayimukeleki nokuthi amamodeli angakakhishwa asevele enekhono kangangokuthi ukuqhubekela phambili ngaphandle kokuqapha okungcono kuzwakala kungenangqondo. Abasebenzi base bezwile kakade inguqulo yalokhu ngaphakathi, ngakho-ke imibono yomphakathi ingabuka kuqala ukuhlanganiswa kunokuqinisekisa isivumelwano esiqediwe.
Ingabe i-OpenAI izohambisana nesithembiso sika-Anthropic sokuhlola okuhlanganisiwe?
Ngemva kwamahora amaningi indaba ka-Amodei, u-Altman ubhale ukuthi i-OpenAI izohambisana nesithembiso somhloli ohlanganisiwe nokufinyelela okufana nokwabasebenzi kanye nemininingwane eminingi maduze. U-Elon Musk ubhale ukuthi uDario uqinisile, ebeka ama-CEO amathathu ayizimbangi ngesiqubulo esifanayo. UHugging Face ucele ukujoyina uhlelo lomhloli futhi wethula uhlelo lwe-Open Alignment Initiative. IYurophu ayikwazi ukunikeza ukuxolelwa okufanayo kwe-US antitrust okushicilelwe yi-Amodei, ngakho-ke ukuxhawulana kuseMelika ikakhulukazi okwamanje.
Kungani izazi zezibalo zigxeka i-OpenAI's Navier-Stokes sprint?
Izazi zezibalo zichaza ukuthi i-ejenti ye-OpenAI engaba yizinkulungwane eziyishumi, egijima amahora angamashumi ayisishiyagalombili nesishiyagalombili iNavier-Stokes njengokuzingela izindebe okubeka phambili ukunqoba kunokuqonda. UTristan Buckmaster uthe i-OpenAI inikeza ukubalwa okungenamkhawulo kanye nokubhala kuphela uma ilahla umuntu osebenzisana ne-Anthropic; wakubiza ngokuthi ukufumbathisa futhi wenqaba. I-OpenAI ithi izikhuthazo zayo zeCodex bezingeke zithinte uhlelo, kodwa isalokhu ingakholwa. U-Andreas Thom usaqhubeka nokugcizelela ukuthi izingxoxo ze-ChatGPT ngomsebenzi wakhe siqu zinikeze intuthuko yemodeli kamuva.
Ingabe i-OpenAI ithole inqubekela phambili kwenye inkinga yeMillennium Prize?
I-OpenAI ithi isivele yenze intuthuko enkulu kwenye inkinga yeMillennium Prize, kodwa ngeke isho ukuthi iyiphi. I-Clay Mathematics Institute isadinga isikhathi eside sokwamukelwa komphakathi, ngakho akekho ophumelele ngokusemthethweni okwamanje. Ngaphakathi kwenkundla, leyo mfihlo kanye nomjaho wangaphambilini weNavier-Stokes kufundeka njengoba amalebhu esevele egijima indebe elandelayo. Izingxoxo zeVerge ziveza ukungezwani njengoba izazi zezibalo zifuna ukuqonda kuyilapho i-OpenAI ifuna ukuwina.