La cusbooneysiiyo maalin kasta1794 sheekooyin la xaqiijiyay

AI News.Sawaxan la'aan.

Isha la hubiyay caynsanaanta AI ee soo saarista alaabada, siyaasada beddelka, cilmi baarista badbaadada, iyo dhaqdhaqaaqa warshadaha, oo ay ku sharaxeen Ingiriis cad koox waxbarasho oo aan faa'iido doon ahayn.

Isha la xaqiijiyay

Sheeko kastaa waxay la xidhiidhaa caddaynta ugu xooggan ee la heli karo: ilaha asalka ah marka la heli karo, haddii kale si cad loo nisbeeyay warbixinta.

Ingiriisi cad

What happened, why it matters, and what to watch — without the jargon.

Buuxiyaha ma jiro

Marka calaamaduhu dhuuban yahay, waxba ma daabacno halkii aan ku dhejin lahayn quudinta.

Kaydka wararka

Kaydka wararka AI — Bogga 141

Source-checked AI stories, newest first, for people who need to understand AI without chasing hype.

Source-provided image accompanying Preprint reports language-model outputs can leak secrets held in context

Sheekada ugu sareysa

Amniga6 min akhri

AmnigaAugust 21, 20266 min akhri

Warbixinta daabacaadda daabacaadda wax-soo-saarka luqadda-qaabka waxay daadin kartaa siraha lagu hayo macnaha guud

Daabacaadda arXiv ayaa sheegaysa in moodooyinka luqaddu ay muujin karaan xogta xasaasiga ah ee lagu meeleeyay macnaha guud, xitaa ka dib marka ay diidaan codsiyada soo saarista tooska ah. Tijaabooyin la kantaroolay, qorayaashu waxay yiraahdeen weerarrada la qabsiga waxay ka soo kabteen siro gaagaaban oo laga helay wax soo saarka caadiga ah oo ay ka soo saareen nambarada Lambarka Bulshada oo buuxa wakiilka qaab-soo-saarka.

Akhri sheekada Isha la xaqiijiyay: arxiv.org

Hal-abuurnimoAugust 21, 20265 min akhri

Thinkingbox benchmark finds AI agents struggle to reliably complete stateful business workflows

A new arXiv paper introduces Thinkingbox, a sandbox and benchmark for testing AI agents on multi-step business tasks. The strongest tested model achieved a 65.36% pass@1 score but only a 25.25% pass^20 score, highlighting the gap between occasional success and dependable execution.

arxiv.org

Hal-abuurnimoAugust 21, 20265 min akhri

Preprint reports asymmetric risks that LLM compression metrics can miss

An arXiv preprint evaluating 3 LLMs across 11 compression methods reports that aggregate accuracy and perplexity can conceal uneven knowledge loss, overconfidence on newly lost information, and offsetting shifts in subgroup bias.

arxiv.org

Sheekooyin badan

9 sheekooyin

Hal warbixin oo faa'iido leh usbuuc kasta

La soco AI adoon ku noolayn quudinta.

Hel wararka AI ee toddobaadka la xaqiijiyay, xogta asalka ah, agabka waxtarka leh, xulashada barashada, iyo shaqooyinka AI ee cusub.

Reach people who are learning AI

Shaqaalaynta xirfadle AI ama soo saarista alaab AI faa'iido leh? Horay dadka halkan u yimid si ay wax u bartaan oo wax u bartaan.

Ku dhaji shaqada AI Soo gudbi qalab AI

AI News.Sawaxan la'aan.

Kaydka wararka AI — Bogga 141

Warbixinta daabacaadda daabacaadda wax-soo-saarka luqadda-qaabka waxay daadin kartaa siraha lagu hayo macnaha guud

Thinkingbox benchmark finds AI agents struggle to reliably complete stateful business workflows

Preprint reports asymmetric risks that LLM compression metrics can miss

Sheekooyin badan

Preprint finds irrelevant text can shift multimodal model judgments predictably

Preprint proposes SingularClip to preserve learning plasticity in changing tasks

Preprint proposes a probe-tested gate for dynamic ensembles under distribution shift

Preprint traces how Llama 3.1 8B models numerical sequence structure

Preprint proposes adaptive feature selection for personalized fall prevention

Preprint proposes ECG model for atrial-fibrillation detection across lead configurations

Preprint proposes physics-guided neural operator for wireless radio maps

Preprint proposes physics-constrained generative AI for global tropical-cyclone forecasts

Preprint reports LLM scheduler cut data-center energy use 32% and waiting time 30%

La soco AI adoon ku noolayn quudinta.

Reach people who are learning AI