Sam Altman says ChatGPT 5 is an expert. Small maths slip shows why it’s not perfect

 OpenAI’s GPT-5, hailed as a PhD-level expert, replaces previous models with enhanced reasoning and coding capabilities. Despite improvements in accuracy and reduced hallucinations, it still commits occasional errors, like a basic decimal subtraction mistake. This highlights the importance of clear instructions for optimal performance, even with advanced AI. OpenAI’s GPT-5, hailed as a PhD-level expert, replaces previous models with enhanced reasoning and coding capabilities. Despite improvements in accuracy and reduced hallucinations, it still commits occasional errors, like a basic decimal subtraction mistake. This highlights the importance of clear instructions for optimal performance, even with advanced AI.