The article critiques OpenAI for limiting web search capabilities in ChatGPT, making it difficult for users to access accurate information. Despite the high cost of LLM inference, the user interface hinders easy search activation and ignores personalization settings. This raises questions about the implementation and costs of web search within the product.