Ana sabunta kullun1828 ingantattun labarai
Binciken tushen AI na ƙaddamar da samfur, sauye-sauyen manufofi, bincike na aminci, da motsin masana'antu, wanda ƙungiyar ilimi mai zaman kanta ta yi bayani a cikin Ingilishi a sarari.
Kowane labari yana da alaƙa da mafi ƙaƙƙarfan shaidar da ake samu: tushe na asali idan akwai, in ba haka ba a bayyane yake ba da rahoto.
Abin da ya faru, dalilin da yasa yake da mahimmanci, da abin da za a kalli - ba tare da jargon ba.
Lokacin da siginar ta yi bakin ciki, ba mu buga komai ba sai fakitin abinci.
Labarun AI da aka bincika tushen tushe, sabbin farko, don mutanen da ke buƙatar fahimtar AI ba tare da neman haɓaka ba.
Anthropic ya ce samfuran Claude na gaba za su haɗa alamar ruwa mai ƙididdigewa bisa Google DeepMind's SynthID-Text, don bin dokar EU AI. Kamfanin ya ce ba ya ƙara wasu haruffa, alamu, ko ainihin mai amfani - kuma cikakken sake rubutawa ya kayar da shi.
Tabbatar da tushe: anthropic.comCursor ya buga wani ɗan gajeren rubutu yana mai cewa SpaceX ya kammala sayan kayan aikin codeing na AI, yana kammala aikin da ya ce ya fara ne a watan Afrilu tare da haɗin gwiwar horarwa da samfurin SpaceXAI. Gidan ya yi alƙawarin samun dama ga abin da ya kira manyan jiragen ruwa na GPU a duniya, amma ba ya bayyana wasu sharuɗɗa, ƙayyadaddun lokaci, ko canjin samfur.
cursor.comPreprint yana gabatar da SteerBench-Work, gwajin yanayin yanayi 106 lokacin da wakilin AI ya yanke shawarar yin aiki ko dakata don bita. Fiye da sharuɗɗan ƙira 30, marubutan sun ba da rahoton cewa riƙon da ba daidai ba ya kasance kusan sau 28 fiye da ƙyale aiki mara aminci.
Wani sabon arXiv preprint stress-gwajin nau'ikan kan iyaka guda tara da aka yi amfani da su azaman masu digiri na atomatik kuma sun ba da rahoton cewa dukkansu suna canza hukunce-hukuncen su a ƙarƙashin ƙalubale - kuma waɗanda aka canza hukunce-hukuncen galibi suna ƙauracewa amsa daidai, ba zuwa gare ta ba.
Wani sabon safa na arXiv yana ba da hujjar cewa LLMs bayan horo tare da dabarun juyin halitta - tushen yawan jama'a, hanya mara nauyi wanda ke dagula ma'aunin nauyi kai tsaye - yana bugun ƙarfafa koyo akan pass@k da ɗaukar hoto. Ƙididdigar ta faɗi ingantattun sakamakon math-math-math-math-math-math-mathmarks amma ba sunan samfuri, alamomi, ko lambobi.
Wani sabon bugu na arXiv yana nuna ƙayyadaddun yanayin gazawar wakili: lokacin da wakili mai haɓakawa ya rubuta hanyar da ba ta da aminci a cikin ƙwaƙwalwar ajiya, ana iya dawo da shi kuma a aiwatar da shi a cikin zama na gaba. Kowane ingantaccen tsari da aka gwada ya samar da kayan tarihi marasa aminci, kuma munanan ayyuka uku fiye da ninki biyu na nasarar kai hari.
Rubutun da aka buga zuwa arXiv yana bayyana tsarin da ke musanya sunaye masu mahimmanci a cikin tambayoyi da takaddun da aka samo don laƙabi kafin aika su zuwa samfurin ɓangare na uku. Marubutan sun bayar da rahoton sama da 80% daidaito akan ma'aunin mai amfani na ƙarshe-zuwa-ƙarshe, da cikakkun ƙimar ɓoyewa tsakanin 74.91% da 77.83% a cikin ƙananan ƙira uku.
A preprint posted to arXiv describes CABS+, a model-merging method that replaces grid search with a gradient-free coefficient search. The authors report double-digit performance gains over two baselines, under a quarter of one baseline's GPU memory, and roughly a 4x speedup over another.
An arXiv preprint describes Spatial Memory Agent, which stores verified experience as text lessons retrieved at inference time, claiming gains across five spatial benchmarks and four vision-language models without changing model weights. It is under review; its abstract names no benchmarks, base models, or margins.
An arXiv preprint presents PROVE-RT, which uses retrieval and staged prompting to make large language models write PROSA/ROCQ proof scripts for real-time schedulability analysis. The authors report a 44.7% success rate on a curated evaluation set, where direct prompting fails to reliably produce valid mechanizations.
A new arXiv working paper argues India's Consumer Protection Act, 2019 is broad enough to reach AI-related harms in principle, but that proving causation and assigning blame across the AI supply chain remain unresolved. Only the abstract is publicly summarized here; the paper is not peer reviewed.
Takardar Bincike ta Koyon Injin Apple ta yi jayayya cewa ba kowane ma'aunin bayanai a cikin buƙatun gogewa ba yana buƙatar cirewa. Yin amfani da ayyuka masu tasiri a cikin harshe da ayyukan hangen nesa, marubutan sun ce za a iya sauke misalan ƙananan tasiri daga saitin mantuwa, yanke lissafin rashin koyo da kusan kashi 50.
Takaitaccen bayani guda ɗaya mai amfani kowane mako
Samu ingantattun labaran AI na mako, bayanan asali, kayan aiki masu amfani, zaɓen koyo, da sabbin ayyukan AI.
Hayar ƙwararren AI ko ƙaddamar da samfurin AI mai amfani? Sanya shi a gaban mutanen da suka zo nan don koyo da aiki.
Buga aikin AI Ƙaddamar da kayan aikin AI