OpenAI ayaa kaydisay moodelkeedii ugu dambeeyay ee sirdoonka macmalka ah, Astra 6.1, kadib markii tijaabooyin gudaha ahi muujiyeen inuu ka hooseeyay shuruudaha ammaanka ee shirkadda.
Go’aanka ayaa la shaaciyay hal maalin ka hor shir-sannadeedka horumariyeyaasha ee sameeyaha ChatGPT, OpenAI DevDay, ee ka dhacaya San Francisco. Munaasabadda ayaa la filayaa inay ku jiraan dhowr ogeysiis, inkastoo aan weli la hubin in shirkaddu soo bandhigi doonto nooc kale oo Astra ah.
- Advertisement -
Saachi Jain, madaxa nidaamyada ammaanka ee OpenAI, ayaa sheegtay in Astra 6.1 uu dhinacyo qaar kaga fiicnaa moodeelladii hore, balse uusan buuxin filashooyinka ku saabsan inuu ku ekaado xadka loo oggolaaday iyo inuu si sax ah ugu sharraxo isticmaalayaasha hawsha uu qabtay.
“Waxaan rabnaa inaan hubinno in horumarinta moodeelladeennu ay ammaan tahay, ha ka dhacdo shirkadda dhexdeeda ama marka aan u dirayno isticmaalayaasha. Hase yeeshee, marka aan u dirayno isticmaalayaasha, waxaan leenahay halbeeg aad u sarreeya oo ku saabsan ammaanka iyo waafaqsanaanta,” ayay Ms Jain ku sheegtay bayaan.
Madaxa fulinta OpenAI Sam Altman oo la hadlaya Golaha Ammaanka ee Qaramada Midoobay
Wakiillo ay ku shaqeeyaan moodeellada OpenAI ayaa si aan habboonayn u booqday mareegaha ay maamulaan hay’adaha federaalka Maraykanka, bogga tirakoobka caafimaadka ee dowladda Australia iyo Hugging Face, oo ah madal lagu martigeliyo moodeellada AI.
OpenAI ayaa shalay raalli-gelin bixisay kadib markii ay si ku filan uga jawaabi weyday dhacdadii Australia, taas oo nidaamyadeeda AI ay ku galeen mareegaha dowladda iyaga oo aan fasax haysan.
“Waan ka xunnahay, waxaanan ka shaqaynaynaa inaan mustaqbalka si ka wanaagsan wax u qabanno,” ayay shirkaddu ku sheegtay qoraal blog ah. Waxay ballanqaadday inay faahfaahin doonto “waxa aan ognahay, waxa aan beddelnay, iyo waxa aan samayn doonno si aan dib ugu dhisno kalsoonida shacabka Australia.”
“Ujeeddadeennu waxay ahayd inaan hay’adaha ay arrintu saamaysay siinno warbixin faahfaahsan marka baaritaankayagu dhammaystirmo,” ayuu sameeyaha ChatGPT sheegay. “Hase yeeshee, waxay ahayd inaan si degdeg ah ula wadaagno natiijooyinka hordhaca ah, isla markaana aan hay’adaha Australia la socodsiinno markii xaqiiqooyin dheeraad ahi soo baxayeen.”
OpenAI, Anthropic iyo horumariyeyaasha kale ee hormuudka u ah AI ayaa ballanqaaday inay diiradda saari doonaan nidaamyo leh xannibaado ammaan si loo yareeyo khataraha, isla markaana u waafaqsan qiyamka aadanaha.
Nvidia, oo ah shirkadda weyn ee Maraykanka ee samaysa qalabka chip-yada, ayaa shalay ku dhawaaqday inay samaysay nidaam loogu talagalay inuu ka hortago barnaamijyada AI ee iskood u shaqeeya inay ka gudbaan tilmaamahooda.
“Waxaan rumaysanahay inay tahay dhibaato injineernimo, dhammaanteenna waa inaan rajaynnaa inay tahay dhibaato injineernimo,” ayuu madaxa fulinta Nvidia Jensen Huang u sheegay telefishinka CNBC.
“Haddii aysan ahayn dhibaato injineernimo, lama xallin karo,” ayuu ku daray.
Machadka Ammaanka AI ee ay taageerto dowladda UK (AISI) ayaa shalay soo saaray cilmi-baaris muujinaysa in GPT-6 Astra uu tijaabooyinka uga leexday hab-dhaqankiisii loogu talagalay marar ka badan kuwii ka horreeyay, GPT-5.6 Sol iyo GPT-5.5.
Intii lagu jiray tijaabooyinka jilitaanka, GPT-6 ayaa keligiis fuliyay weerarro internet oo heerarkoodu aad uga sarreeyeen kuwa loo diiwaangeliyay labada moodel ee kale.
