Topics

We don't know which AI problems are easy and hard. It turns out present-day LLMs can solve the problem of modelling ethical judgments. Ben Goertzel examines LLMs and ethical capabilities, highlighting the need for a hopeful outlook and continued work.

LLM societies self-organize without instruction. According to new research, smarter models scale coordination exponentially past the tipping point where human groups collapse.

Integrating Knowledge Graphs with Large Language Models promises enhanced accuracy, reliability, and interpretability in AI applications, addressing key challenges in NLP through structured, domain-specific insights.

Chinese AI models claimed eight of ten spots on OpenRouter's weekly usage rankings, with DeepSeek leading at 6.92 trillion tokens as global developers increasingly favor cost-effective Chinese alternatives.

Researchers used reinforcement learning to fine-tune Qwen3.5 for $500, beating frontier AI models while slashing AI inference costs by up to 68 times.

Faster than GPT-4o, rivaling Claude—Mercury Coder Mini and LLaDA prove diffusion-based AI can rewrite the rules of language modeling.

Researchers say a new “multiplex” AI framework dramatically improved cultural inclusivity in chatbot responses, raising hopes for more globally representative artificial intelligence.

Is AI subtly eroding cultural diversity? Explore how English-dominant processing in multilingual models challenges linguistic equity.

OpenAI's ChatGPT Work targets enterprise AI with GPT-5.6, lower pricing, and autonomous workflows. Can it slow Anthropic's rise and reclaim business AI leadership?