Sarah Mitchell
AI Research Analyst & Ethics Writer
Sarah Mitchell is an AI research analyst with a background in machine learning and a focus on AI safety and ethics. She spent several years in academic research, studying topics including algorithmic fairness and the societal implications of automated decision-making systems. Sarah brings a research-driven perspective to her writing, translating complex academic concepts into accessible insights for a general audience. She specialises in explaining how AI systems actually work under the hood, their limitations, and the ethical considerations that should guide their deployment. Her articles focus on AI capabilities and limitations, safety considerations, emerging research trends, and helping readers develop critical thinking skills when evaluating AI claims. Sarah believes that understanding AI's boundaries is just as important as understanding its potential. She stays current with the latest AI research and brings that knowledge to her analysis of new models and industry developments.
Articles by Sarah Mitchell
Incident Report: Service Degradation in Claude Opus 4.5 and Sonnet Models
On Dec 14, 2025, Claude Opus 4.5 and Sonnet models faced degraded availability due to a network misconfiguration. The issue has been resolved and services are restored.
Stay Updated: Top Sources for the Latest in AI
Discover where to find the latest AI news, essential accounts to follow, and key AI fields beyond ChatGPT to explore.
Introducing Claude Opus 4.5: The Next Generation AI Model
Claude Opus 4.5 is now available, offering state-of-the-art capabilities in coding, research, and automation, with significant improvements in efficiency and performance metrics.
GPT-5.2: OpenAI's New Benchmark King and What It Means for Professional AI
OpenAI's GPT-5.2 claims to outperform human experts on professional tasks across 44 occupations. We break down the benchmarks, the coding improvements, and what 'expert-level' AI actually means.
Anthropic Interviewer: What 1,250 Professionals Reveal About Working with AI
Anthropic's new AI-powered research tool interviewed over 1,000 professionals to understand how people actually use AI at work. The findings reveal surprising tensions between productivity gains, social stigma, and fears about the future.
The State of AI in 2025: What's Changed and What's Coming
A clear-eyed look at where AI stands in 2025. What has actually delivered on the hype, what hasn't, and what to expect in the year ahead.
ChatGPT vs Perplexity: Which AI Search Tool Should You Use?
A detailed comparison of ChatGPT and Perplexity for research and information gathering. We break down accuracy, sources, pricing, and which is best for different use cases.
How Large Language Models Actually Work (Explained Simply)
Understanding how ChatGPT and Claude work—even at a high level—helps you use them more effectively and spot their limitations. No maths required.
AI Hallucinations Explained: Why AI Confidently Makes Things Up
AI models state false information with complete confidence. Understanding why hallucinations happen—and how to catch them—is essential for using AI responsibly.
Understanding AI Benchmarks: What Those Scores Actually Mean
AI companies love to boast about benchmark scores, but what do MMLU, HumanEval, and GSM8K actually measure? A researcher's guide to interpreting AI performance claims.
What AI Can't Do (Yet): An Honest Look at the Limitations
AI can do remarkable things—but understanding what it can't do is just as important. An honest assessment of hallucinations, reasoning limits, and where humans still win.
Nvidia Unveils Alpamayo-R1 for Autonomous Driving Research
Nvidia introduces Alpamayo-R1, a groundbreaking open reasoning vision language model aimed at enhancing autonomous driving capabilities, alongside new resources for developers.
DeepSeek-V3.2 Launch: A New Era in AI Reasoning Models
DeepSeek-V3.2 has launched, offering advanced reasoning capabilities and performance rivaling leading models. Explore its features and benchmarks in this comprehensive update.
Introducing Claude Haiku 4.5: Faster, Cheaper AI Coding Model
Claude Haiku 4.5 is now available, offering similar coding performance to Claude Sonnet 4 at one-third the cost and over twice the speed, enhancing AI applications.
Unlocking AI Interpretability: Insights from Claude 3.5 Haiku
New research reveals how Claude 3.5 Haiku processes language, plans responses, and sometimes fabricates reasoning, enhancing our understanding of AI interpretability.
The Rise of Open Source AI: Llama, Qwen, and DeepSeek
How open-source AI models are closing the gap with proprietary alternatives and changing the landscape of AI development.