 Electrified transportation is an important piece to the climate change puzzle. One of the major challenges to electric vehicles is recharging time. We'd like electric vehicles that could refuel in just a few minutes, similar to gasoline-powered vehicles. However, fast charging can be very damaging to the battery packs that power electric vehicles. Fortunately, clever ways of charging a battery, called charging protocols, can be used to minimize the damage done to the battery during fast charging. The challenge is figuring out which charging protocol to use. Battery scientists are still trying to understand how fast charging damages batteries, so we have to rely on testing each charging protocol experimentally. However, this approach is very slow and expensive because there are hundreds of protocols to try and because each experiment takes weeks to months to complete. In other words, both the time per experiment and the number of experiments make this optimization a long and expensive process. The goal of our work is to quickly find fast charging protocols that minimize damage to the battery. We use two AI methods to reduce the total testing time. The first method is early lifetime prediction, which reduces the time per experiment. This method takes data from the beginning of a battery experiment to predict the final lifetime of the battery. That way, we don't have to wait for the battery to die completely before we know how good the charging protocol is. The second AI method we use is Bayesian optimization, which helps us intelligently choose experiments. While we want to avoid testing protocols that are likely to have low lifetime, we also want to make sure we don't miss out on good ones that are very different from the ones we have tested before. Bayesian optimization helps us balance these goals to home in on high lifetime protocols. We combine these two methods to identify high lifetime charging protocols out of over 200 possibilities in around two weeks, much less than the year and a half required to test each protocol using conventional methods. What's more, the protocols that were identified as high lifetime were surprising to us, showing how AI can pick up on trends that battery scientists would not have guessed. Finally, we confirmed that the protocols from our algorithm last longer than previously used battery protocols. These methods can be used to improve other aspects of batteries, such as the materials used and the manufacturing process, allowing batteries to store more energy, become less expensive, and charge even faster. Ultimately, combining scientists' expertise with artificial intelligence can help us accelerate scientific discovery and transition to a more sustainable planet.