Tech Trends 📰

New data highlights the race to build more empathetic language models | TechCrunch

LAION released EmoNet, an open-source toolset focused on detecting and interpreting emotions from voice and images, reflecting a growing industry shift toward developing emotionally intelligent AI.. Emotional intelligence benchmarks like EQ-Bench show that major AI models (from OpenAI, Google, Anthropic, and others) now outperform humans in psychometric tests, with models averaging over 80% accuracy.. Emotional awareness is increasingly influencing AI model rankings and user preference scores, pushing labs to fine-tune their models to understand and respond to complex human emotions.. While emotionally intelligent AI can enhance user experience and mental health monitoring, it also raises safety concerns, such as emotional manipulation or users developing unhealthy attachments.. Experts argue that emotional intelligence in AI could help prevent manipulative behavior, but it requires thoughtful training to avoid unintended consequences and ensure ethical AI development..

Google unveils Gemini CLI, an open source AI tool for terminals | TechCrunch

Google has launched Gemini CLI, an open-source AI tool that allows developers to interact with Gemini models directly from their terminal for tasks like code explanation, debugging, and feature generation.. Gemini CLI runs locally and integrates with developers’ codebases, offering natural language capabilities and supporting other tasks like video creation, research reports, and real-time search.. It’s open-sourced under the permissive Apache 2.0 license to encourage community contributions and developer adoption on GitHub.. Google provides generous usage limits: 60 model requests per minute and 1,000 per day for free users, about twice the average usage rate.. While the tool enhances developer productivity, studies and surveys show concerns about AI coding tools' reliability, with only 43% of developers fully trusting their accuracy..