The author shares their personal experiences using AI tools, primarily focusing on large language models (LLMs) like ChatGPT and GitHub Copilot. They highlight coding as the area most transformed by AI, emphasizing their extensive use of Copilot since its launch. The autocomplete feature significantly enhances productivity by predicting code lines and suggesting idiomatic patterns, making it an invaluable tool. The author notes that while Copilot excels at common tasks, it struggles with complex algorithms and requires human oversight to validate its outputs. They find the agent mode somewhat useful but note it can lead to errors, particularly in unconventional scenarios. In research and search, the author finds LLMs useful for casual queries and retrieving specific information, like identifying articles or books. They appreciate the ability to ask for explanations of well-known concepts, which sometimes yield better results than traditional documentation. However, they express skepticism about the reliability of LLMs, especially for product searches and literature reviews. The author warns against over-relying on AI for research, as it can produce outdated or incorrect information. While LLMs can aid in finding sources, they lack the critical perspective of human experts, making their outputs potentially misleading.