There are thousands of chemicals, both manmade and naturally occurring, to which humans are exposed. It is now common knowledge that a drug's potency is only one factor in determining its success. Absorption, distribution, metabolism, excretion, and toxicity are only some of the features that need to be taken into account. Due to the extensive resources needed to evaluate a chemical, little is learned about its potential toxicity. There are a wide range of toxicities or adverse medication effects that need to be assessed during both the preclinical as well as clinical trial phases to ensure patient safety. To determine the safety of certain chemicals, scientists have typically used in in vitro as well as in vivo trials. Yet, not only are such research costly and time-consuming, but animal testing experiments specifically are increasingly being criticized for being unethical. The field has seen some success employing conventional machine learning (ML) techniques. When combined with Big Data and AI, machine learning's successes in fields like NLP, speech recognition, image identification, combinatorial chemistry, and genomics suggest it may be useful for toxicity prediction in the modern era. In this piece, we will apply state-of-the-art machine learning techniques to the problem of toxicity prediction. These techniques include deep learning, regression trees, k-nearest neighbors, and support vector machines, among others. We also talk about how changing the machine learning algorithm's input parameters, such as moving from a focus on chemical structure description alone to additionally including study of human transcriptome data, can significantly improve its prediction accuracy.