Chatbots can guide measurement of absorption spectra with improved quality

Absorption spectroscopy is a long-standing instrumental method for qualitative and quantitative analyses, yet is prone to errors given the richness of sample type, suite of instrumental limitations, and scope of possible human errors; hence, guidance for proper on-the-fly measurements is essential. Recent advances in transformer-based large language models have yielded conversational artificial intelligence (AI) chatbots. Here, the capabilities of six major chatbots (ChatGPT 4o, Gemini Advanced 1.5 Pro, Microsoft Copilot, Meta AI, Claude 3.5 Sonnet, and Grok 2) have been examined by requesting practical advice to achieve improved quality in the measurement of absorption spectra, where verbal descriptions were provided concerning the features of a poor-quality, experimentally collected spectrum already in hand. The suggestions by the chatbots were surprisingly on point in most cases. The results here imply a valuable role for chatbots in support of human analysts as well as automated experimental systems during the measurement of absorption spectra.