Automated Machine Learning (AutoML) can dynamically design suitable model architectures and hyperparameters for various data and tasks, making it easier to use machine learning models. However, designing appropriate model frameworks and hyperparameters is only a small aspect of a data scientist's responsibilities. Task formulation, result recommendation, and visualization are not addressed by existing AutoML frameworks. To bridge this research gap, we draw inspiration from the recent success of Large Language Models (LLMs) and aim to integrate AutoML systems with LLMs, enabling everyone to use machine learning models with ease and become a data scientist.