🏛️ Uhulumeni uzokwakha ilabhorethri entsha ukuze i-UK iqhubeke nokuthuthukisa ubuchwepheshe be-AI ↗
I-UK isungula i-Fundamental AI Research Lab esekelwa nguhulumeni, iwubeka njengomsebenzi “wesibhakabhaka esiluhlaza” - uhlobo oluyingozi, oluhamba kancane, futhi ngezinye izikhathi olukhokhelayo ngendlela eyenza wonke umuntu abukeke sengathi ulele. (GOV.UK)
Ukugxila akukhona nje "amamodeli amakhulu, ama-GPU amaningi" - kubhekana namaphutha aqhubekayo njengokubona izinto ezingekho, inkumbulo emfushane, kanye nokucabanga okungalindelekile, kanye nokunikeza abacwaningi ukufinyelela ekubaleni okungathi sína nge-AI Research Resource. Kuzwakala kunengqondo kakhulu... futhi, buthule, njengomzamo wokuvimba izingqondo ezinhle kakhulu zase-UK ukuthi zingashiyiwe kwenye indawo ngokushesha. (GOV.UK)
🧨 I-CEO ye-Nvidia iphakamisa ukuphela kokutshalwa kwezimali ku-OpenAI, Anthropic ↗
UJensen Huang ukhombisa ukuthi iNvidia ingase ingaqhubeki nokutshala imali kuma-lab e-AI asemngceleni ngendlela efanayo - njengoba i-IPO dynamics (kanye nobukhulu bokuhlola obuxoxwayo) okwenza kube nzima ukufeza leso sitayela sokuxhasa ngezimali. (Reuters)
Kuyinguquko yethoni efanele ukubhekwa njalo: I-Nvidia iyinkosi enekhono lokukhuphula wonke lo mnotho, kodwa iphakamisa ukuthi "ukuba nezingxenye zabavukuzi" akuseyona into edlalwayo ngaso sonke isikhathi. Noma mhlawumbe kumane kuwukuzivikela ngokuzwakalayo, okuyinto eyenziwa yi-CEO, njengokuphefumula. (Reuters)
🧩 Okukhethekile: Iqembu elikhulu lobuchwepheshe lisekela i-Anthropic empini yePentagon njengoba abatshalizimali bezama ukunciphisa ukungqubuzana ngezivikelo ze-AI ↗
Ukungezwani kwe-Anthropic nePentagon sekuphenduka isitofu esicindezelayo - kubikwa ukuthi abatshalizimali bafuna izinga lokushisa lehle, kuyilapho inkampani izama ukugcina ulaka lwayo ngolimi lokuvikela (ikakhulukazi mayelana nokuqapha). (Reuters)
Umbhalo ongaphansi kwendaba ucishe ube mkhulu kunombhalo: enkathini ye-AI, amagama enkontileka awayona “into engalungile ngokomthetho,” empeleni kuyinqubomgomo yomkhiqizo - futhi inquma ukuthi imodeli iba ithuluzi, isikhali, noma umthwalo wemfanelo omkhulu. (Reuters)
🪖 USam Altman uyavuma ukuthi i-OpenAI ayikwazi ukulawula ukusetshenziswa kwe-AI yiPentagon ↗
Kubikwa ukuthi u-Altman utshele abasebenzi be-OpenAI ukuthi abakwazi ukulawula indlela iPentagon esebenzisa ngayo i-AI yayo uma isithunyelwe - okuyinto efika ngesivinini esikhulu ngoba ichaza ukwesaba abantu abebelokhu bekuzwa. (The Guardian)
Isizinda esibanzi siwukwanda kokungqubuzana phakathi kokuthi “sizosiza, ngemithetho” kanye nokuthi “sizosiza, ukuyeka ngokuphelele,” kanye nokubuyiselwa emuva kwangaphakathi nomphakathi lapho ukwamukelwa kwezempi kuzwakala kuphuthunyiswa noma kunethuba. Izimiso zokuziphatha lapha aziyona imigqa ecocekile kodwa ziwukuchitheka kopende okumanzi - wonke umuntu uyangena, bese ephikisana ngokuthi isicathulo sikabani. (The Guardian)
🧬 Ubudlelwano obusha be-AI ku-genomics ne-sanger institute kanye ne-Google DeepMind ↗
I-Wellcome Sanger Institute yethula ubudlelwano bezemfundo obuxhaswe yi-DeepMind obugxile ekusebenziseni i-AI ku-genomics - obubekwe njengendawo yokuqala yohlobo lwayo yomuntu we-DeepMind kule ndawo ethile. (sanger.ac.uk)
Okuthakazelisayo (futhi, ngokusobala, okuqabulayo kancane) ukugcizelela ezinkingeni ze-genomics ezingakahlolwa kahle lapho i-AI ingekho khona yonke indawo - kanye nephuzu elicacile lokuthi i-DeepMind ayiqondisi ucwaningo lwalowo muntu. Kufana nokunikeza umuntu irokhethi bese uthi “hamba uyothola okuthile,” kunokuba “hamba uyolungisa imephu yethu yendlela.” (sanger.ac.uk)
Imibuzo Evame Ukubuzwa
Iyini i-Fundamental AI Research Lab esekelwa nguhulumeni wase-UK, futhi izokwenzani?
I-Fundamental AI Research Lab esekelwa nguhulumeni ibekwa njengomzamo wokucwaninga "wesibhakabhaka esiluhlaza" - umsebenzi onobungozi obukhulu ongathatha isikhathi ukuwufeza. Esikhundleni sokugxila kuphela ekukhuliseni amamodeli amakhulu, ihlose ukubhekana nezinkinga eziqhubekayo ezifana nokubona izinto ezingekho, inkumbulo emfushane, kanye nokucabanga okungalindelekile. Iphuzu liwukuthi impumelelo ivela ezintweni eziyisisekelo, hhayi nje ekungezeni ama-GPU amaningi.
I-UK Basic AI Research Lab ingabasiza kanjani abacwaningi ukuthi bafinyelele ekubalweni okujulile?
Kanye ne-UK Fundamental AI Research Lab, uhlelo luqokomisa ukufinyelela ekubalweni okukhulu nge-AI Research Resource. Empeleni, lokho kuvame ukusho ukuthi abacwaningi bangenza izivivinyo ebezizovinjelwa yizindleko noma ingqalasizinda. Kuphinde kuvumele amaqembu ukuthi ahlole imibono ngezinga lapho izinkinga ezifana nokuthembeka nokuqina ziba khona, hhayi nje ngokwemfundiso.
Kungani i-UK igcizelela ukungaboni kahle, inkumbulo emfushane, kanye nokucabanga okungalindelekile?
Lobo buthakathaka yibo obuvela ekusetshenzisweni futhi bungaqeda ukwethembana ngokushesha. Ukugxila okukhulunywe ngakho kusikisela ukuthi umgomo awukona nje ikhono, kodwa ukuthembeka - ukunciphisa imiphumela eyenziwe, ukuthuthukisa indlela amamodeli aphatha ngayo umongo omude, nokwenza ukucabanga kungaguquguquki. Lolo hlobo lomsebenzi luvame ukuhamba kancane futhi lube yingozi, yingakho lubhekwa njengocwaningo oluyisisekelo.
Ukushintsha kwezwi likaNvidia ekutshalweni kwezimali ku-OpenAI noma ku-Anthropic kukhombani ngempela?
Umbiko uwubeka njengesibonelo sokuthi iNvidia ingase ingaqhubeki nokutshala imali kuma-lab asemngceleni ngendlela efanayo, ikakhulukazi njengoba i-IPO dynamics kanye nosayizi omkhulu wamasheke kwenza lelo qhinga libe nzima. Ngisho noma umholi "wokukhetha nokushaya" kwihadiwe ye-AI, kusikisela ukuthi ubunikazi bezabelo abuyona into engcono kakhulu ngaso sonke isikhathi. Kungaba futhi imiyalezo eqaphile, okuyinto evamile kumazwana aphezulu.
Kungani impikiswano ye-Anthropic yePentagon mayelana “nolimi lokuvikela” iyindaba enkulu kangaka?
Iphuzu elibalulekile lalesi sihloko ukuthi amagama enkontileka angaba inqubomgomo yomkhiqizo - ikakhulukazi uma ethinta ukuqapha kanye nokunye ukusetshenziswa okubucayi. Kubikwa ukuthi abatshalizimali bafuna ukunciphisa ukungqubuzana, kuyilapho inkampani izama ukugcina umkhawulo wayo ezivikelweni. Ezimweni eziningi zokusetshenziswa kwe-AI, lezo zigaba zibumba ukuthi uhlelo lungasetshenziselwa ini, nokuthi yiziphi izingozi inkampani ezamukela ngempumelelo.
Kusho ukuthini uma uSam Altman ethi i-OpenAI ayikwazi ukulawula indlela iPentagon esebenzisa ngayo i-AI?
Ichaza umkhawulo osebenzayo: uma ithuluzi selisetshenzisiwe, unjiniyela wokuqala angase abe nekhono elilinganiselwe lokulawula ukusetshenziswa okulandelayo. Lokho kufinyelela kakhulu ngoba kukhomba ukwesaba okuyinhloko abantu abakuphakamisayo mayelana nokwamukelwa kwezempi - imithetho ingase ibe khona esigabeni senkontileka, kodwa ukuphoqelelwa kungaba nzima. Kuphinde kubonise ukungezwani okukhulu phakathi "kosizo, ngemingcele" kanye "nosizo, kungakhathaliseki."