Tokenization Challenges Hinder Multilingual LLMs
A 2026 study highlights tokenization's role in poor performance of low-resource language models, with 340+ languages struggling due to suboptimal segmentation.
Archive
Browse all published articles.
ARCHIVE
-
-
Github Copilot Shifts to Token-Based Billing Model
Github Copilot is switching to a token-based billing model starting June 1, potentially increasing costs for smaller developers by up to 25 times.
-
Microsoft and Nvidia Team Up on AI PCs Using Actual Agents
Microsoft and Nvidia are collaborating on AI PCs that run actual agents, not Copilot, with devices expected to debut next week at Computex and Build.
-
HuggingFace Introduces Token-In, Token-Out for RL Training
HuggingFace outlines a method to maintain token integrity during reinforcement learning, ensuring gradients are computed on the exact tokens the model generated. This approach resolves issues with token mismatch during multi-turn interactions.
-
SoftBank to Invest Up to €75 Billion in French Data Centers
SoftBank plans to invest up to €75 billion to build data centers in France, aiming to add 5 gigawatts of capacity by 2031.
-
Anthropic Study Reveals Men Use AI Coding Agents More Than Twice as Often as Women in Social Science Research
Anthropic's study found men use AI coding agents more than twice as often as women in social science research, with economists leading at 39% and education researchers at 4%.
-
SoftBank Plans 75 Billion Euro AI Data Center Buildout in France
SoftBank announced a 75 billion euro investment in AI data centers in France, with 3.1 gigawatts of capacity planned by 2031.
-
AI Search Agents Rely on Memory, Not Web Research
A study reveals leading AI search agents often confirm existing knowledge rather than research the web, with models like GPT-5.4 and Kimi-K2.6 scoring high on static benchmarks.
-
TechCrunch Publishes AI Glossary to Explain Complex Terms
TechCrunch has released an updated AI glossary to clarify terms like AGI and LLMs, offering a living document to help users understand the evolving field.
-
Equity Podcast Explores AI-Pilled Companies and Workforce Impact
Equity podcast hosts examine AI's role in layoffs and workplace tensions as 2026 tech layoffs nearly match 2025 totals.
-
Groq Raises $650M to Expand Inference Cloud Business
AI chip startup Groq is reportedly raising $650 million to scale its inference cloud business, following a $20 billion not-acqui-hire deal with Nvidia in December.
-
Cognition's Devin AI Coding Agent Secures $1 Billion Raise
Cognition, maker of AI coding agent Devin, raised $1 billion at a $26 billion valuation, emphasizing human-AI collaboration over replacement.