Generative AI is rapidly changing how higher education assesses learning: models now draft feedback on student writing, assign scores, and even explain their decisions. But does a system's resemblance to human judgment make it trustworthy? This book argues that it does not, treating assessment in the age of AI not as an accuracy problem to be solved by larger models, but as a human-centered socio-technical practice in which feedback must be designed, scoring must be governed, and claims must be ...