Model Alignment

The world has achieved a major breakthrough in AI research by designing models with 1 quadrillion parameters. This is considered the holy grail of Deep Learning. However, as the researchers test the models, they realize that the machine is biased, making it incapable of making a moral judgement. This leads to the creation of an AI-powered detective, who must find a way to align the model's moral compass by investigating the source of the biased data.