Résumé : Historically, researchers have focused on analyzing WEIRD, adult perspectives on technology. This means we may not have technology developed appropriately for children and those from non-WEIRD countries. In this paper, we analyze children and parents from various countries' perspectives on an emerging technology: conversational agents. We aim to better understand participants' trust of agents, partner models, and their ideas of "ideal future agents" such that researchers can better design for these users. Additionally, we empower children and parents to program their own agents through educational workshops, and present changes in perceptions as participants create and learn about agents. Results from the study (n=49) included how children felt agents were significantly more human-like, warm, and dependable than parents did, how participants trusted agents more than parents or friends for correct information, how children described their ideal agents as being more artificial than human-like than parents did, and how children tended to focus more on fun features, approachable/friendly features and addressing concerns through agent design than parents did, among other results. We also discuss potential agent design implications of the results, including how designers may be able to best foster appropriate levels of trust towards agents by focusing on designing agents' competence and predictability indicators, as well as increasing transparency in terms of agents' information sources.