#mlresearch — Public Fediverse posts

Google deploys Gemini 3.5 Flash to Search, Spark, and enterprise as Pro delays to June. Meta cuts 8,000 roles while shifting 7,000 to AI work. Anthropic hires Andrej Karpathy for pretraining research. The infrastructure shift is underway—compute becoming the constraint, not capability.

#AI #LaborMarkets #MLResearch

https://www.implicator.ai/google-ships-agents-meta-cuts-people-anthropic-buys-brains/

Research shows adding more features to ML regression models can introduce hidden structural risks. Every additional feature creates dependencies on upstream data pipelines, and low-signal variables may appear important due to noise, leading to models that behave inconsistently when deployed. https://www.marktechpost.com/2026/03/08/beyond-accuracy-quantifying-the-production-fragility-caused-by-excessive-redundant-and-low-signal-features-in-regression/ #AIagent #AI #GenAI #MLResearch #MarkTechPost

#Reproducibility #OpenScience #ComputationalSocialScience #WebData #DigitalBehavior #MLResearch #NLPResearch #SocialMediaData

#CallForPapers: Our full-day workshop at The Web Conference 2026 #TheWebConf2026 invites submissions on reproducible and reusable computational approaches for social and web data.
👉 https://easychair.org/cfp/r2cass2026

Deadline: Dec 18th, 2025

Language Models are Few-Shot Learners, page 33 mentions bidirectional models as a promising direction. Any recent movement in that direction? #ai #mlresearch