Recent artificial intelligence (AI) models trained on vast amounts of data have proven remarkably powerful. For instance, textual and visual content generated by such models may occasionally appear to rival human creativity. Yet, in this talk, I will argue that notable gaps remain between current AI models and intelligent systems that more reliably serve human needs. Apart from explainable AI, which provides a rationale for each decision, real-world systems ought to more robustly cope with diverse kinds of data and possess mechanisms that allow us to control their operation. This in turn also requires them to acquire a deeper understanding of human language and cognition, and ultimately a broader understanding of the world.
Please register via: https://hpi.de/en/the-hpi/registration/2022/antrittsvorlesung-prof-de-melo/
The event will take place at the Hasso Plattner Institut, lecture hall 1, at 5 pm.
