Farkon Ƙoƙarin Sabon Dabaru don Haɓaka AI Don Gano Kimiyya
Farkon Ƙoƙarin Sabon Dabaru don Haɓaka AI Don Gano Kimiyya
Haɓaka basirar ɗan adam don gano kimiyya ya zama babban buri ga ƙwararrun ƙwararrun fasaha. Kamfanoni kamar OpenAI da Anthropic sun sami dubun-dubatar biliyoyin kuɗi tare da alkawuran ci gaban AI a cikin magani, ilmin halitta, da kimiyyar lissafi. Koyaya, binciken kimiyya na gaskiya wanda AI ke jagoranta ya kasance mai wuyar gaske, kamar yadda abubuwan da suka faru a baya suka nuna kamar binciken lissafi na ChatGPT da ba a bayyana ba. Babban ƙalubalen, a cewar masana, shine manyan samfuran harshe na yanzu (LLMs) ba su da ƙarfin gaske don samar da sabon ilimin kimiyya da kansa.
Me yasa Big AI Labs ke gwagwarmaya tare da Gano Kimiyya
Markus Buehler, farfesa injiniyan MIT, ya gano ƙayyadaddun ƙayyadaddun ƙayyadaddun ƙayyadaddun ci gaba na AI na yau. Ya bayar da hujjar cewa tsarin samar da wutar lantarki daga OpenAI da Anthropic ba a tsara su don gano gaskiya ba. Gine-ginen su ya dogara ne akan ƙirar ƙira daga bayanan data kasance, ba akan ƙirƙirar sabbin dabaru ko hasashe ba.
An kwatanta wannan a cikin faɗuwar ƙarshe lokacin da aka yi watsi da binciken da ChatGPT ya yi da sauri. Lamarin ya nuna tazarar da ke tsakanin ikon nazari na AI da kerawa, tunanin ganowa. Kalubale ne mai tunawa da sauran yunƙurin AI inda fasahar ke fama da asali, kamar sukar da AI ‘dan wasan kwaikwayo Tilly Norwood ya fuskanta na rashin kerawa na gaske.
Babban Matsala tare da Model AI na Yanzu
Manya-manyan ƙirar harshe sun yi fice wajen sarrafa bayanai da sarrafa bayanai. Suna iya taƙaita rubutu, amsa tambayoyi, har ma da rubuta lamba bisa bayanan horon su. Koyaya, suna aiki a cikin iyakokin abin da suka rigaya suka koya.
Binciken kimiyya, ta yanayinsa, yana buƙatar shiga cikin abin da ba a sani ba. Ya ƙunshi ƙulla sabbin alaƙa tsakanin fagage daban-daban da ba da shawarar ra'ayoyin waɗanda ba su cikin kowane tsarin bayanan horo. Wannan tsalle-tsalle ne wanda AI na yanzu, mai da hankali kan ƙirƙirar abun ciki da sarrafa kansa, ba a gina shi don yin ba. Masana'antar tana haɓakawa, kamar yadda aka gani tare da ci gaba kamar sabuntawar WordPress Gutenberg da ke shimfida aikin bugu na AI, amma babban ƙalubalen ganowa ya rage.
Gabatar da Labs Mara Hankali: Sabuwar Hanyar zuwa AI don Kimiyya
Don magance wannan gibin, Farfesa Buehler ya haɗu da Labs marasa ma'ana tare da Yuan Cao, tsohon babban masanin kimiyyar bincike a Google DeepMind. Farawa yana da nufin yin majagaba na asali daban-daban don haɓaka AI don binciken kimiyya. Maimakon dogaro da ɗimbin bayanai kawai, suna gina tsarin da ke da ikon yin tunani a tsakani.
Labs marasa ma'ana kwanan nan sun sami dala miliyan 13.5 a wani zagaye na tallafi wanda Playground Global ke jagoranta. Zagayen ya ga sa hannu daga AIX Ventures, E14 Fund, da MS&AD Ventures. Wannan gagarumin saka hannun jari yana jaddada imanin kasuwa ga sabbin dabarunsu.
Koyo daga lokacin "Aha" a cikin Tarihin Kimiyya
Hasashen Buehler shine yawancin manyan bincike sun taso daga lokacin "aha". Waɗannan lokuta ne inda masanin kimiyya ya yi amfani da ka'ida ko ra'ayi daga wani fanni don magance matsala a cikin wani yanki na daban. Wannan giciye-pollination na ra'ayoyi mabuɗin ga nasara.
Misali na yau da kullun shine aikin John Hopfield a cikin 1982. Ya yi amfani da ra'ayoyi daga ilimin kimiyyar kwayoyin halitta zuwa fagen hankali na wucin gadi. Wannan ya haifar da haɓaka hanyoyin sadarwar Hopfield, nau'in hanyar sadarwa na jijiyoyi masu iya koyo da tuno abubuwan tunawa. Tunani ne na juyin juya hali da aka haife shi daga haɗa lamuran da ba su da alaƙa.
Yadda Labs 'AI mara hankali ya bambanta da na yau da kullun
AI da ake haɓakawa a Labs marasa ma'ana an ƙirƙira shi don kwaikwayi wannan ƙarfin ɗan adam don fahimtar tsaka-tsaki. Manufar su ba shine ƙirƙirar samfurin harshe mafi girma ba amma don gina tsarin da zai iya yin tunani a cikin sassan kimiyya.
Zane-zanen Ilimin Tsare-tsare: Maimakon horarwa akan rubutu kaɗai, AI nasu yana haɗa ingantaccen ilimi daga fannonin kimiyya da yawa, daga ilmin halitta zuwa kimiyyar lissafi. Injin Hulɗar Halittu: Babban fasaha yana mai da hankali kan gano kwatankwacinsu da kamanceceniya tsakanin ra'ayoyin da ba su da alaƙa, babban direban ƙirƙira ƙima. Hasashen Hasashen: Ana ƙirƙira tsarin don ba da shawarar hasashen kimiyya da za a iya gwadawa, ba wai kawai nazarin bayanan da ke akwai ba.
Wannan hanya tana wakiltar babban ficewa daga dabarun sayan mafi girmakamfanonin fasaha, irin su Zendesk saye na AI farawa Forethought, wanda sau da yawa mayar da hankali kan tace data kasance abokin ciniki aikace-aikace maimakon majagaba sabon nau'i na gano.
Makomar Ganowar AI-Driven
Idan an yi nasara, fasahar Labs marasa ma'ana na iya haɓaka bincike a wurare masu mahimmanci. Ka yi tunanin AI wanda zai iya ba da shawarar sabon fili na miyagun ƙwayoyi ta hanyar haɗa ka'idoji daga sunadarai da kwayoyin halitta. Ko samfurin da ke ba da sabon abu don ɗorewar makamashi ta hanyar haɗa ra'ayoyi daga nanotechnology da thermodynamics.
Abubuwan yuwuwar aikace-aikacen suna da yawa, daga haɓaka binciken likita zuwa warware ƙalubalen muhalli masu rikitarwa. Wannan yana wakiltar gaba na gaba don AI, yana motsawa fiye da aiki da kai don zama abokin tarayya na gaskiya a cikin basirar ɗan adam.
Kammalawa: Na gaba Wave na AI Innovation
tseren don haɓaka AI don binciken kimiyya yana ƙaruwa, amma nasara ta gaskiya na iya kasancewa tare da ƙwararrun farawa kamar Labs marasa ma'ana. Mayar da hankalinsu kan tunani na tsaka-tsaki yana ba da kyakkyawar hanya mai ban sha'awa fiye da iyakoki na manyan harsunan yanzu. Tafiya don ƙirƙirar AI wanda zai iya ganowa da gaske yana farawa.
Kasance da sabuntawa akan sabbin sabbin abubuwa a cikin AI da fasaha. Don ƙarin fahimta da kuma raba wannan labarin cikin sauƙi, ƙirƙirar shafin haɗin yanar gizon ku na kyauta akan abin da ba shi da kyau don tsara abubuwan da kuka fi so.
