Not only that, but scientists in all disciplines are skilled professionals. Designing tools which are effective will depend on understanding the nature of their expertise. This raises all kinds of questions: are automated number-crunching tools that index, search and sort the way forward? Do we need other kinds of tools that model and highlight patterns, trends and anomalies in complex data and structures? To what extent do computerbased tools need to reveal and be explicit about their underlying assumptions and constraints? And as tools become more complex and work on ever greater datasets, it may be difficult to know when they malfunction, or when they are misapplied. Another concern is how such tools represent complexity and make it tractable, whether it be modelling the earth’s support systems or the human immune system. If a computer simulates a complex system, does it simply create a new one that needs further analysis and understanding? How can the ensuing knowledge be communicated and acted upon to solve problems in the world? For example, how can the results of computational analyses from many millions of data points be represented in meaningful ways? As we take on more complex problems, use more sophisticated models, and rely on increasingly powerful computing resources and vast quantities of data, these issues will become more significant. The ability to provide increasingly sophisticated tools to augment our human capabilities speaks strongly to the human values associated with our desire for productivity and industriousness in our lives, and our aspirations for greater knowledge. We will need to fathom out how best to represent and present information. This involves working out how to make data from all kinds of different sources intelligible, usable and useful. These may come from research labs, but equally may come from an ever-growing stream of data from the increasing array of sensors placed throughout the world. It also entails figuring out how to integrate and replay, in meaningful and powerful ways, the masses of digital recordings that are being gathered and archived, such that professionals and researchers can perform new forms of computation and problem-solving, leading to novel insights. Questions for interaction and design Is further automation the way forward for augmenting human thinking and problem-solving? How can the interaction and design of new computational tools be structured so they do not impede creative engagement? What new toolkits can be developed to enable scientists, and others to create tools for themselves to solve their own problems and explore new avenues? Questions of broader impact What will such tools mean for the nature of expertise in future? Will scientists become too dependent on tools? If so, what does this mean for the nature of invention and discovery? Will computer-based tools eventually become so complex they can no longer be understood by the people who developed them?