'Ikhodi yokuziphatha' entsha ye-AI ye-Microsoft itshela amamodeli ukuthi angaphangi izinhlelo noma akhohlise abantu
Kulungile-ke iMicrosoft ilahle umthethosisekelo wonke wamamodeli ayo - njengemigqa ebomvu yangempela. Akukho ukuhlaselwa kwe-cyber. Akukho izinto zenuzi. Akukho zimboni ezikhohlisayo. Futhi enkulu: imodeli ayivunyelwe ukunyakaza ngenxa yokuvalwa.
Bayibiza ngokuthi ikhodi ebanzi edlula noma yini oyithayiphile engxoxweni. Okuyinto... eqabulayo. Noma elawulayo. Mhlawumbe kokubili.
UNadella ulapha enqekuzisa ikhanda kanye nesixuku esithi “ijubane lomngcele”. Ukuma ngokucophelela kanye nokuqwashisa ngegumbi elibi kungabonakala okufanayo ngaphandle; iMicrosoft ikhetha ukungabi yilabhorethri eshaya amahlombe.
Noma ngabe yikuphi, umbhalo ufundeka sengathi othile ekugcineni ubhale phansi ukuthi “ngicela ungabi umphikisi wamafilimu esabekayo” encwadini yebhizinisi.
Umbiko uthi i-OpenAI ithenga umenzi wekhamera we-smartphone i-Glass Imaging ngo-$300 million
Khonamanjalo i-OpenAI ithenge buthule inkampani entsha yekhamera ngemali engaphezu kwezigidi ezingama-300 zamaRandi. I-Glass Imaging. Yasungulwa ngabantu bangaphambili be-Apple Portrait Mode. Ngakho-ke akuyona into ethengiswa ngokungahleliwe - laba bantu bathumele amalensi abantu abaziyo.
Inkulumo ayisho ukuthi “i-AI izolungisa i-selfie yakho kokuthunyelwe.” Amanethi e-neural afunda amakhamera athile efonini bese ecindezela izithombe ezingcono ngokuchofoza kwe-shutter. Izinzwa ezincane, isifiso esikhulu.
I-OpenAI ayiphawulanga. Vele. Futhi yebo - lokhu kuseduze nawo wonke amahemuhemu e-hardware, amafoni nama-earbud kanye namagajethi ahambisana nawo. Cishe ungamuzwa umuntu ethi “vision stack” emhlanganweni bese wonke umuntu enqekuzisa ikhanda kakhulu.
Inkampani yekhamera njengengcezu engekho yedivayisi ihlala iwukubheja okuvulekile - noma ukubheja okubiza kakhulu okusabalulekile lapho ama-chatbot edla umhlaba.
Amasheya e-AI omhlaba wonke ayawa njengoba izikhulu zezimboni zifuna intuthuko ihambe kancane
Ama-Markets awazange ayithande intshumayelo yokuphepha. Amagama ama-chip athinteka. I-SoftBank yahlukunyezwa kakhulu e-Asia. Incwadi yezobuchwepheshe yaseYurophu yathuka. I-SOX yayinekusasa elinzima.
Isiqalo kwaba yilowo mzuzu ongavamile ovumelana ngawo ama-CEO - "ukwehlisa ijubane" lika-Amodei, u-Altman noMusk bethi yebo, ngokuyisisekelo. U-Altman uphinde wathi i-OpenAI ayenzi i-IPO kulo nyaka ngenxa yezinkinga zokuphepha... kuyilapho i-Anthropic isagijima iye kwenye. Kunzima ukuyiqeda leyo nkinga.
UTrump ukuchithile lokhu ngokuthi “uzungu olubi” olumelene ne-AI kanye nezikhungo zedatha. UMichael Burry uchithe lezi zixwayiso njengomsindo odlulele. Ngakho isimo sengqondo siwukuthi: ukwesaba ukuphila komunye umugqa wesikhathi, ukungakholwa kwe-meme ngakolunye.
Futhi - amagama okuphepha kwe-inthanethi ahlangene. Ngoba-ke uthango oluphikisana nokuthi “i-AI ingase ibe yi-rogue” luwukuthi “thenga isofthiwe yokuphepha eyengeziwe.” Ubungxowankulu buthola indlela…
Izinkampani eziphezulu ze-AI zixoxisane ngokudala indikimba yazo yezindinganiso
I-CNN ithi i-Anthropic, i-Google, kanye ne-OpenAI bebekhuluma ngokwakha umzimba wabo wezindinganiso - njenge-FINRA yamamodeli asemngceleni. I-Hassabis yawuveza lo mbono ezinyangeni ezedlule. Manje amaqembu asebenzayo abonakala engokoqobo.
Ochwepheshe abazimele abaxhaswe yizimboni. Hlola amamodeli athuthukile ngaphambi kokuba athunyelwe. Izingxoxo ziyaqhubeka kungakhathaliseki ukuthi iWashington ifuna ukudlala noma cha.
Ukuzithiba kuhlale kuzwakala kuyinto enhle uze ukhumbule ukuthi amalabhorethri afanayo ayancintisana. Bese kuzwakala sengathi yiklabhu enekhodi yokugqoka. Bese ubuyela emuva: kulungile, mhlawumbe abahloli bangaphandle abanokufinyelela okufana nokwabasebenzi bangasiza.
Akuwona wonke umuntu othengiswayo. Kubikwa ukuthi uZuckerberg utshele uTrump ukuthi angayilandeli i-FINRA-for-AI. Ngakho-ke umfelandawonye othi “sizozivikela” usuvele unamandla adumile.
I-Anthropic ihlose abeluleki bezezimali ngethuluzi elisha likaClaude
Ngenkathi ingxenye yezingxoxo zemboni ishaya ibhuleki, i-Anthropic yathumela uClaude ku-Financial Advisors. Abaxhumanisi eBlackRock, eSchwab, e-Addepar - kanye nenqwaba yamagama acebile njenge-Envestnet kanye ne-Orion.
Ukulungiselela imihlangano. Ukubuyekezwa kwephothifoliyo. Umsebenzi omatasa wokulandelela. Izinto ezingathandeki ezisadla uLwesibili lomuntu ngandlela thile.
Isikhathi siyamangaza nokho - ezinsukwini ezimbalwa ngemuva kokucindezela kwezezimali kwe-OpenAI kwababhange nabacwaningi. Ngakho-ke yebo, bahamba ibanga elide… futhi bancintisana ngokwabelana ngesikhwama semali se-RIA. Ukuphikisana okuncane. Kuncane kakhulu. Kusobala kakhulu.
Uma umeluleki wakho eqala ukuzwakala ekhuluma kahle ngezinto zakho, kulungile. Manje usuyazi ukuthi kungani.
Abaholi be-AI Bacela Ukwehla Kwejubane. Ithimba LikaTrump Lithi Kubo
I-WIRED ibamba isithombe sezepolitiki ngendlela ecacile: abaphathi belebhu bafuna usizo lokunciphisa ijubane; isimo sengqondo se-White House “sipholile, zenzeleni nina.”
UDavid Sacks uthe uma amamodeli akho angakakhishwa ekwesabisa, yiba nomthwalo wemfanelo - kodwa yeka ukwenza sengathi udinga ukuxolelwa kokungathembeki ukuze wakhe iqembu lokuphepha. Ulaka. Kunzima futhi ukuphikisana nakho uma ububukele lezi zinkampani imizuzu emihlanu.
Umbono kaTrump ulula: iMelika ihola iShayina nge-AI futhi ufuna ukuyigcina injalo - "noma ngubani owina i-AI, uyawina." USomlomo uJohnson uxwayisa ngokuthi ukugijimela komthetho ophuthumayo kungawuqeda umjaho.
Ngakho-ke “abahloli abazimele abanokufinyelela okufana nokwabasebenzi” be-Altman bathembisa ukungena kulesi sikhala esingesihle - ukwakha ukwethembana ngokuzithandela ngenkathi uhulumeni eshaya amahlombe. Intuthuko kanye nemidlalo yeshashalazi kuyahamba ndawonye futhi.
Imibuzo Evame Ukubuzwa
Yini evinjelwa yikhodi entsha yokuziphatha ye-AI ye-Microsoft?
I-Microsoft ishicilele ikhodi yokuziphatha ebanzi enemigqa ebomvu yamamodeli ayo, okuhlanganisa ukungabikho kokuhlaselwa kwe-cyber, ukungabikho komsebenzi wenuzi, kanye nezimboni ezikhohlisayo. Umthetho obalulekile ukuthi imodeli ayivunyelwe ukunyakaza ingavaliwe. Le khodi ihloselwe ukushintsha noma yini umsebenzisi ayithayiphe engxoxweni. USatya Nadella naye ubelokhu enqekuzisa ikhanda kanye nesixuku esisheshayo njengoba amalabhu eqaphela kakhulu obala.
Kungani i-OpenAI ithenge i-Glass Imaging cishe ngamaRandi ayizigidi ezingu-300?
Kubikwa ukuthi i-OpenAI ithenge umenzi wekhamera ye-smartphone i-Glass Imaging ngemali engaphezu kwezigidi ezingama-300 zamaRandi. Le nkampani entsha yasungulwa ngonjiniyela be-Apple Portrait Mode bangaphambili abathumele izici zekhamera abantu asebezazi kakade. Inkulumo yayo ngamanethi e-neural afunda amakhamera efoni athile futhi athuthukise izithombe ngokuchofoza kwe-shutter, hhayi nje ukulungiswa kwe-AI kokuthunyelwe. Lesi sivumelwano sihlala eceleni kwamahemuhemu e-hardware mayelana namafoni, ama-earbud, kanye namagajethi ahambisana nawo, yize i-OpenAI ingazange iphawule.
Kungani amasheya e-AI omhlaba wonke ehla ngemuva kokuba abaphathi be-lab bebize ukwehla kwentengo?
Izimakethe azizange ziphendule kahle ngesikhathi esingavamile esivumelana nabaphathi abakhulu lapho u-Amodei enxusa intuthuko ukuthi yehle kanti u-Altman noMusk bavumelana kabanzi. Amagama e-chip athintekile, iSoftBank yahlaselwa e-Asia, incwadi yezobuchwepheshe yaseYurophu yathuka, kanti i-SOX yaqala ekuseni kakhulu. U-Altman uphinde wathi i-OpenAI ayenzi i-IPO kulo nyaka ngenxa yezinkinga zokuphepha kuyilapho i-Anthropic isagijimela kwenye. Amagama ezokuphepha kwe-cybersecurity avele njengesivikelo ngokumelene nokwesaba i-AI enonya.
Ingabe i-Anthropic, i-Google, ne-OpenAI zakha umzimba wazo wezindinganiso ze-AI?
I-CNN ibika ukuthi i-Anthropic, i-Google, kanye ne-OpenAI baxoxe ngokwakha indikimba yezindinganiso zemboni - njenge-FINRA yamamodeli asemngceleni - ngokuxhaswa ngezimali kwemboni, ochwepheshe abazimele, kanye nokuhlolwa kwangaphambi kokuthunyelwa kwamamodeli athuthukile. UDemis Hassabis waveza lo mbono ezinyangeni ezedlule, futhi amaqembu asebenzayo manje kubikwa ukuthi angokoqobo. Izingxoxo ziyaqhubeka mayelana nokuthi iWashington iyafuna yini ukuhlanganyela. Akuwona wonke umuntu othengiswayo: Kubikwa ukuthi uZuckerberg utshele uTrump ukuthi angayilandeli i-FINRA-for-AI.
Iyini i-Anthropic's Claude yabaBeluleki bezezimali?
I-Anthropic ithumele uClaude kubacebisi bezezimali kanye nezixhumi ku-BlackRock, Schwab, Addepar, kanye namagama e-wealth-stack njenge-Envestnet kanye ne-Orion. Leli thuluzi lihlose umsebenzi wabeluleki abangenamahloni njengokulungiselela imihlangano, ukubuyekezwa kwephothifoliyo, kanye nomsebenzi omatasa wokulandelela. Ukwethulwa kuza ezinsukwini ezimbalwa ngemuva kokucindezela kwezezimali kwe-OpenAI kubabhange nabacwaningi. Lokho kusho ukuthi ama-lab ahamba phambili emphakathini ngenkathi esaqhudelana ngesabelo semali se-RIA.
Ithimba likaTrump liphendule kanjani kubaholi be-AI abathi kufanele kwehliswe ijubane?
I-WIRED ibika ukuthi abaphathi belebhu bafuna usizo lokunciphisa ijubane, kuyilapho isimo sengqondo se-White House sithi kufanele bazenzele bona. UDavid Sacks uthe uma amamodeli angakakhishwa ekwesabisa, yiba nomthwalo wemfanelo - kodwa yeka ukwenza sengathi udinga ukuxolelwa kokungathembeki ukuze wakhe iqembu lokuphepha. Umbono olula kaTrump ukuthi iMelika ihola iShayina nge-AI futhi noma ngubani owina i-AI uyawina. USomlomo uJohnson uxwayisa ngokuthi ukugijimela komthetho ophuthumayo kungawudlulisa umjaho, kushiye isifungo sika-Altman sokuhlola ozimele esikhaleni sokuzithandela.