The Archive

Mark Wahlberg joins Bruce K. Lee at Disrupt to discuss investing, entrepreneurship, healthcare, wellness and building businesses.

A new feature coming to Slack will allow you to build interactive reports, polls, dashboards, presentations, microsites, and other tools directly inside a chat. With Slackforce Surfaces, you can describe to Slackbot what you need, and it will use AI to gather information from relevant conversations and connected apps, like Google Drive or Salesforce, to create it. Once Slackbot creates a Surface, you can share it with colleagues and pin it to channels, allowing other people to view it, interact with it, and leave comments. In one example shared by Slack, a user asks Slackbot for help creating...

Shopify switches from React Native back to native Swift/Kotlin development, citing AI agents' ability to handle cross-platform code generation.

The company said Pro subscriptions put the most strain on its systems, so it's pausing sign-ups while adding more capacity.

A new report released Thursday by Anthropic alleges persistent distillation attacks by China-based AI companies, which have escalated in recent months as competition in the space has intensified.

Meta's newest app Muse is off to a slower start than the company's other apps, like Meta AI or Threads.

It's the hot new thing in tech, and it's where all the jobs are. Students who don't learn to use it fall behind. And to help them catch up in time, its creators are graciously providing the resources and curriculum for learning it, often pro bono. That's the narrative AI companies are pitching schools on right now, but according to New York Times education technology reporter Natasha Singer, it's also how the tech industry has built its influence in classrooms for the past 15 years. In a new book, Coding Kids, Singer details how tech companies took a central role in promoting computers and co...

"Bankruptcy cannot become the new land grab for AI.”

GPU-CFR achieves 80x speedup for counterfactual regret minimization via static dataflow compilation and CUDA graph replay, addressing GPU kernel dispatch bottleneck.

Proposes γ*-concept shift framework using entropic optimal transport to unify covariate and concept shift bounds, bridging theory-practice gap in distribution shift generalization.

Empirical study showing Mixture-of-Experts models overfit more severely to repeated training data than dense Transformers across 80M-1B scale.

Evaluates edge-deployable 2–8B vision-language models (Qwen3-VL, Gemma3) versus BioCLIP on species identification from camera-trap imagery.

Framework for measuring causal business impact of generative engine optimization and sponsored placement visibility in AI-generated answers.

Studies distance generalization in Transformers—whether positional encoding helps when inter-token distances differ between training and inference.

Proposes artificial id mechanism—adaptive internal drive for agentic systems to autonomously decide task continuation, stopping, and behavioral transitions without external specification.

Commentary on translating AI principles to protocols and standards (EU AI Act, NIST AI RMF, ISO 42001) as institutional repair mechanisms.

Modular four-stage guitar tablature transcription system (TART) handling expressive techniques, string-fret accuracy, and noisy real-world audio.

Come inside the mind of a bot trying to convince the internet it's human.

MindTopo benchmark evaluates foundation-model topological reasoning (continuity, separation, order, enclosure, knots) grounded in cognitive science.

CausalArena benchmark standardizes causal discovery evaluation across graph families and mechanisms, addressing fragmentation as foundation models enter the field.

3D Gaussian splatting adapted for mmWave radar novel view synthesis with physics-faithful rendering and phase information preservation.

Nuha-Speech introduces 1.5M-sample Arabic speech-QA corpus and infrastructure for training multilingual speech-LLMs in underrepresented language.

CoRA-NAS combines zero-cost proxy ranking with learning-curve extrapolation for neural architecture search using 1% validation cost.

Multi-signal hallucination detection pipeline achieves F1=0.915 on HaluEval using DeBERTa-v3, MC Dropout, and temperature calibration.

AssayBench-Loop benchmark for adaptive experimental design in CRISPR screening spans 1,389 screens across five phenotype categories.

Analysis unifies linear recommendation models under nuclear-norm and Frobenius-norm regularization frameworks, explaining performance convergence.

Pocket FM uses AI to produce 99% of its new content, helping make content production about 80 times cheaper.

Roadmap for genuine recursive self-improvement in AI systems progressing from execution autonomy through meta-improvement, using Headroom-Closed Index.

Empirical study evaluates eight time-series foundation models on continuous glucose monitoring forecasting across diabetes populations with multimodal context.

DeBERTa trained on 10M words with visually grounded token embeddings shows ostensive learning persists through training but remains invisible to syntax benchmarks.