imudojuiwọn ojoojumọ1792 daju itan
Agbegbe AI ti a ṣayẹwo orisun ti awọn ifilọlẹ ọja, awọn iyipada eto imulo, iwadii aabo, ati awọn gbigbe ile-iṣẹ, ṣalaye ni ede Gẹẹsi nipasẹ ẹgbẹ ẹkọ ti kii ṣe èrè.
Gbogbo itan ni asopọ si ẹri ti o lagbara julọ ti o wa: awọn orisun atilẹba nigbati o wa, bibẹkọ ti o han gbangba.
What happened, why it matters, and what to watch — without the jargon.
Nigbati awọn ifihan agbara jẹ tinrin, a jade ohunkohun kuku ju òwú kikọ sii.
Source-checked AI stories, newest first, for people who need to understand AI without chasing hype.
A position paper proposes the Transition Complexity Profile, a standardized way to describe how unpredictable and long-range the dynamics of game environments are for game-world modeling and reinforcement learning.
A new arXiv benchmark places 15 language-model agents in a 20-year football-management simulation, testing whether they can make consistent decisions when short-term choices affect long-term outcomes.
A new arXiv benchmark reports that changing only the retrieval method raised a fixed model’s exact accuracy on financial reconciliation cases from 2.05% to 72.44%. The study also finds that a correct root-cause label often does not mean the system returned sufficient evidence for an auditable diagnosis.
Iwadi arXiv tuntun ṣafihan awọn adanwo ofin iwọn-si-aworan fun awọn awoṣe itankale ọrọ-si-aworan kọja awọn isuna iṣiro lati 10 ^ 19 si 10 ^ 22 FLOPs. Awọn onkọwe rẹ ṣe ijabọ pe awọn awoṣe aworan nilo data diẹ sii fun paramita ju awọn awoṣe ede lati ṣe ikẹkọ daradara.
Iwe arXiv tuntun ṣe apejuwe ọna EEG ti ara ẹni ti o ṣe ijabọ deede 92.4% ti o ṣe iyatọ arun Alzheimer lati awọn iṣakoso deede lori ẹgbẹ ADFTD.
Iwe arXiv ṣafihan Agentic ESOpt, ilana ilana itankalẹ ti a dabaa fun ṣiṣatunṣe awọn aṣoju awoṣe ede gigun pẹlu iranti GPU ipele-inference. Awọn onkọwe ṣe ijabọ awọn anfani fun Qwen-3.5-27B lori WebArena-Lite ati awọn ilọsiwaju ni 28 ti awọn eto iṣapeye iyara 36.
Iwe arXiv jiyan pe aṣiṣe onigun mẹrin le ṣe aiṣedeede awọn asọtẹlẹ akoko-akoko alaibamu ati dabaa iṣiro akoko ti o ni idanwo kọja sintetiki, ologbele-sintetiki ati awọn data gidi-aye mẹjọ.
Iwe arXiv ṣafihan Gbigbe ibatan, wiwọn ti bi awọn apẹẹrẹ ikẹkọ ni awọn ipele iṣoro oriṣiriṣi ṣe ni ipa lori ara wọn, ati pe o dabaa Gbigbe-mọ Iṣapẹẹrẹ Iṣapẹẹrẹ Agbara fun ṣiṣatunṣe awọn iṣeto ikẹkọ lakoko ikẹkọ.
An arXiv preprint ṣe apejuwe Co-RL, ilana imudaniloju ọpọlọpọ-oluranlowo ninu eyiti awọn awoṣe ọtọtọ san ẹsan fun ara wọn. Awọn onkọwe ṣe ijabọ awọn anfani kọja ọrọ-nikan ati awọn itọkasi multimodal laisi awọn aami otitọ ilẹ, lakoko ti o jẹwọ awọn eewu ti awọn aṣiṣe ti ara ẹni ati isubu ikẹkọ.
Iwe arXiv tuntun dabaa Iṣẹ-ṣiṣe Iṣẹ-ṣiṣe Itanran-Tuning lati pin lopin itanran-tuning kọja awọn agbegbe iṣẹ-ṣiṣe ti o ni ibatan. O ṣe ijabọ agbegbe iṣẹ-ṣiṣe ti o dara julọ ni iṣapeye, iṣakoso, ati awọn adanwo LLM, ṣugbọn ko fun awọn abajade nọmba tabi ifọwọsi ominira.
Ti a gba si Apejọ Asilomar 2026, iwe kan dabaa SURE-Ridge, ọna fọọmu pipade fun iṣiro ifẹsẹmulẹ-aworan ni awọn awoṣe Gaussian ti o dọgbadọgba. Awọn onkọwe ṣe ijabọ aṣiṣe igbekale kekere kekere ati akoko idanwo ti o yara julo; orisun naa ko fi idi iṣẹ mulẹ kọja awọn ero wọnyi.
Awọn oniwadi ṣe apejuwe J-Miner, ọna kan ti o ṣe iyipada awọn ifihan agbara inu lati awọn classifiers awoṣe ede ti o dara julọ sinu awọn ofin ti a le ṣayẹwo, ṣe ijabọ to 98.3% ẹda ti awọn ipinnu orisun ati pe o fẹrẹ to deede iṣẹ-ṣiṣe kanna ni awọn awoṣe ọmọ ile-iwe ti o kere julọ.
Alaye ti o wulo ni ọsẹ kọọkan
Gba awọn iroyin AI ti a ṣayẹwo ti ọsẹ naa, data atilẹba, awọn irinṣẹ to wulo, awọn yiyan ẹkọ, ati awọn iṣẹ AI tuntun.
Igbanisise ọjọgbọn AI kan tabi ṣe ifilọlẹ ọja AI ti o wulo? Fi si iwaju awọn eniyan ti o wa nibi lati kọ ẹkọ ati ṣe.
Firanṣẹ iṣẹ AI kan Fi ohun elo AI silẹ