For the advancement of Artificial Intelligence, custom hardware solutions using GPUs (graphics processing units) have proven superior in performance than the use of traditional CPUs, as well as another approach used with the use of ASICs that are optimized hardware solutions for specific applications, such as for Deep Learning processing.
The main obstacle to the mainstream use of Artificial Intelligence is the difficulty of eliminating the biases in the results generated in the algorithms used, which are trained in data that in its origin often also contain these vices, which ends up creating an automation of decisions that can impact people's lives.
People generally are not even aware of these facts and with the new data protection laws in the world, an Artificial Intelligence professional who wants to stand out should be concerned with first being able to make "clear" what premises and factors were considered to generate an outcome of the machine learning algorithm that probably had a result in the "real" world.
In other words, the ability of a human operator to be able to explain and interpret the results generated by this algorithm in the context of the problem, I believe it will be a preponderant factor for its use to pass to be disseminated for example in areas that deal with sensitive data such as Healthcare.