🤝 I-Anthropic ingena esivumelwaneni sesikhungo sedatha se-SpaceX njengoba iqhubeka nokubhala ikhodi ye-AI ↗
I-Anthropic ithole isivumelwano esikhulu sokubala ne-SpaceX, yathola ukufinyelela ku-Colossus 1 eMemphis. Lokho kunikeza uClaude amandla amaningi, ikakhulukazi kubasebenzisi be-Claude Code abebelokhu bebhekene nezindonga zokufunwa.
Isici esingajwayelekile? I-SpaceX ithi i-Anthropic nayo inesithakazelo kumandla okubala i-AI ye-orbital. Kuzwakala njenge-sci-fi, mhlawumbe kuyi-toaster encane emafwini, kodwa umjaho wokubala ngokusobala uya ngokuya uba yinyoka.
🧠 I-Anthropic ithi ifundise ama-AI indlela "yokuphupha" ↗
I-Anthropic yethule isici esibizwa ngokuthi “ukuphupha” ngama-ejenti e-AI. Umqondo uwukuthi ama-ejenti abukeze umsebenzi wawo wangaphambilini phakathi kwezikhathi, aqaphele amaphethini, futhi athuthukise indlela aziphatha ngayo ngesikhathi esizayo.
Iqala njengokuhlola kuqala ucwaningo, ngakho-ke akuwona wonke umuntu othola ukufinyelela okwamanje. Noma kunjalo, isiqondiso sicacile - I-Anthropic ifuna ama-ejenti azizwa sengathi awafani namafasitela engxoxo kodwa azizwa sengathi asebenza nabo abazimela kancane. Kuwusizo, kuyamangaza, mhlawumbe kokubili.
🔎 Izifinyezo zokusesha ze-AI ze-Google manje zizocaphuna i-Reddit ↗
I-Google ingeza imibono eminingi emisha ku-AI Mode kanye ne-AI Overviews, okuhlanganisa i-Reddit, amaforamu, kanye nokuthunyelwe kwezenhlalo. Empeleni, ihlanganisa i-hack endala ethi "engeza i-Reddit ekusesheni kwakho" ekusesheni kwe-AI uqobo.
Kuphinde kwengeze umongo owengeziwe mayelana nezixhumanisi, njengamagama abadali namagama omphakathi. Kuhle ngomqondo, yize ukuvumela i-AI ifingqe izinkundla ze-inthanethi kusazwakala njengokucela i-raccoon ukuthi ihlele i-pantry.
⚡ I-Microsoft's AI data center push iyashayisana nemigomo yayo yamandla ahlanzekile ↗
Kubikwa ukuthi iMicrosoft iphikisana ngokuthi ngabe ukwakheka kwayo okunamandla kwesikhungo sedatha se-AI kwenza kube nzima ukuhlangabezana nezinhloso zayo zamandla ahlanzekile. Ingcindezi iwukulinganisa amandla ahlanzekile ngehora, okuyinto eqinile kunokulinganisa kwaminyaka yonke.
Leyo indaba yengqalasizinda ye-AI exakile nge-miniature: wonke umuntu ufuna amamodeli engeziwe, ama-GPU engeziwe, umthamo owengeziwe - bese kungena ibhili kagesi ligqoke amabhuzu ensimbi.
💾 Ukwanda kwe-AI kunyuse i-Samsung ku-$1T ↗
I-Samsung idlule inani lama- $1 trillion njengoba isidingo se-AI chip siqhubeka nokukhanyisa izimakethe zememori. Imemori ye-bandwidth ephezulu ingumlingiswa oyinhloko lapha, ngoba izinhlelo ze-AI zanamuhla ziyayifuna kakhulu.
Amasheya enkampani akhuphuke kakhulu, futhi umyalezo obanzi ulula: ukwanda kwe-AI akugcini nje ngokuvuza amalebhu amamodeli. Kuphakamisa lonke uchungechunge lokuhlinzekwa kwama-chip, ngokungalingani kodwa ngamandla.
🦾 Izinkampani ezisafufusa zaseFrance zembula imodeli ye-AI yamarobhothi nezandla ezifana nezomuntu ↗
I-Genesis AI yembule i-GENE-26.5, imodeli eyakhelwe ukulawula izinhlobo ezahlukene zamarobhothi, kanye nesandla serobhothi esihlakaniphile esenzelwe ukulingisa ukunyakaza kwabantu. Ama-demo ayehlanganisa imisebenzi ebucayi njengokusika utamatisi nokuxazulula i-Rubik's Cube.
Le nkampani entsha ihlose umsebenzi wezimboni kwezezimoto, izinto zikagesi, ikhemisi kanye nezokuthutha. Iphinde iqoqe idatha yabasebenzi abasemsebenzini ngamagilavu esenzwa, ahlakaniphile - futhi amnandi kancane emphakathini.
🧵 I-Corning ibambisana ne-Nvidia ukwandisa umkhiqizo we-fiber optic wase-US ukuze kukhule i-AI ↗
UCorning noNvidia basebenzisana ukuze bandise ukukhiqizwa kwemikhiqizo yokuxhumana kwe-optical e-US yezikhungo zedatha ze-AI. Ama-chips athola ukugqanyiswa, kodwa i-fiber iwumthombo othule wokudlulisa idatha phakathi kwalabo baprosesa.
I-Corning iphinde yaphakamisa imigomo yayo yokuthengisa yesikhathi eside njengoba isidingo sengqalasizinda ye-AI siqhubeka nokusabalala ngale kwama-GPU. Uchungechunge lokuhlinzekwa kwe-AI luphenduka ipayipi lengadi elinamakhanda amaningi, ngokuyisisekelo.
Imibuzo Evame Ukubuzwa
Kungani izinkampani ze-AI zenza amadili engeziwe esikhungo sedatha?
Izinkampani ze-AI zidinga umthamo omkhulu wokubala ukuze ziqeqeshe amamodeli, zikhonze abasebenzisi, futhi zisekele imikhiqizo efana neyama-ejenti. Izipesheli ezifana nokufinyelela kwe-Anthropic ku-Colossus 1 ye-SpaceX zibonisa ukuthi ingqalasizinda ephakathi isibe kanjani. Njengoba isidingo sikhuphuka, ikakhulukazi ngamathuluzi okubhala ikhodi njenge-Claude Code, umthamo owandisiwe wesikhungo sedatha ungasiza ekunciphiseni izithiyo zokusebenzisa futhi ugcine izinsizakalo zisebenza kahle.
Sisho ukuthini isici "sokuphupha" sika-Anthropic kuma-ejenti e-AI?
Isici sika-Anthropic esithi “amaphupho” sichazwa njengendlela yokuthi ama-ejenti e-AI abukeze umsebenzi odlule phakathi kwezikhathi. Umgomo uwukuba ama-ejenti abone amaphethini, afunde ekuziphatheni kwasekuqaleni, futhi enze kangcono ngesikhathi esizayo. Njengoba siqaliswa njengokuhlola kuqala ucwaningo, kusesenesikhathi, kodwa sikhomba kuma-ejenti e-AI ukuthi abe maphikelela kakhulu, aqaphele umongo, futhi agxile emsebenzini.
Kungani i-Google ingeza i-Reddit kanye nokuqukethwe kweforamu ekusesheni kwe-AI?
I-Google izama ukwenza izifinyezo zokusesha nge-AI zifake imibono evela ezindaweni ezifana ne-Reddit, amaforamu, kanye nokuthunyelwe komphakathi. Lokhu kubonisa ukuthi bangaki abantu asebevele besesha ngokungeza i-"Reddit" emibuzweni lapho befuna izimpendulo zesipiliyoni noma zomphakathi. Inselele ukwenza lezo zifinyezo zibe usizo ngaphandle kokunciphisa izingxoxo ze-inthanethi ezibanzi zibe izimpendulo ezilula kakhulu.
Ingqalasizinda ye-AI iyithinta kanjani imigomo yamandla ahlanzekile?
Ingqalasizinda ye-AI ingadinga ugesi omningi, ikakhulukazi njengoba izinkampani zakha izikhungo zedatha ezengeziwe zamamodeli athuthukile kanye nezinsizakalo zamafu. Ukungezwani okubikiwe yi-Microsoft mayelana nokuqhathaniswa kwamandla ahlanzekile ngehora kubonisa ukuthi kungani lokhu kunzima. Ukuqhathaniswa kwamandla ahlanzekile ihora nehora kuqinile kunokuqhathaniswa kwaminyaka yonke, ngakho-ke ukukhula okusheshayo kwe-AI kungenza kube nzima ukwanelisa imigomo yokuzinza.
Kungani inkumbulo ye-bandwidth ephezulu ibalulekile ekukhuleni kwe-AI?
Imemori ye-bandwidth ephezulu ibalulekile ngoba izinhlelo ze-AI zanamuhla zidinga ukuhambisa nokucubungula inani elikhulu ledatha ngesivinini esikhulu kakhulu. Ukwenyuka kwe-Samsung kudlula inani lama- $1 trillion kubonisa indlela isidingo sama-chip e-AI kanye nememori esingazuzisa ngayo abahlinzeki ngale kwamalebhu emodeli uqobo. Emipayipini eminingi ye-AI, ukusebenza kwememori kungaba yinto ebaluleke kakhulu njengamandla ekhompyutha aluhlaza.
Amamodeli e-AI angashintsha kanjani amarobhothi nomsebenzi wezimboni?
Amamodeli e-AI e-robotics ahlose ukusiza imishini ukuthi yenze imisebenzi ehlukahlukene yomzimba ngokuguquguquka okukhulu. Imodeli ye-GENE-26.5 ka-Genesis AI kanye nesandla se-robotic esifana nesomuntu yizibonelo zalolu hlangothi, kanye nemiboniso ehilela ukunyakaza okubucayi njengokusika utamatisi nokuxazulula i-Rubik's Cube. Ukusetshenziswa kwezimboni kungafaka phakathi izimoto, izinto zikagesi, imithi, kanye ne-logistics, ikakhulukazi lapho umsebenzi wezandla ophindaphindwayo uvame khona.