The Practical Applications Of Artificial Intelligence | Seminar Follow By A Panel Discussion

AI technologies already support many everyday products and services, with further applications to come. Professor Marcus Du Sautoy, mathematician, author, and science communicator speak to leading artificial intelligence experts for a discussion and open forum on technologies with AI at their heart to have the power to change the world.

  • Understanding The Weather And Climate
  • Improving Diagnostics And Healthcare
  • Interacting With Computers And Devices By Voice Recognition
  • Supporting Advances In Science

Julia Slingo FRS Chair, Cabot Institute, University of Bristol
Antonio Criminisi Principal Researcher, Microsoft Research
Steve Young Professor of Information Engineering, University of Cambridge, and Senior Member of Technical Staff – Siri Team, Apple
Suzanne Aigrain Professor of Astrophysics, University of Oxford)

Understanding The Weather And Climate
To understand daily weather patterns, meteorologists use AI techniques and simulations that model the Earth’s atmosphere, ocean, and land surface, drawing from a combination of the fundamental laws of physics, complex maths, and a large amount of data about previous patterns in order to make predictions about the future. Julia Slingo FRS (Chair, Cabot Institute, University of Bristol) uses advanced statistics to understand the weather. As the predictive power of such techniques improves, such analyses can also help inform policy debates about the likelihood of extreme weather events, such as flooding.

Improving Diagnostics And Healthcare
One area of significant excitement around the application of AI is in healthcare, and Antonio Criminisi (Principal Researcher, Microsoft Research) is helping drive the development of such applications. One of the most difficult tasks for doctors treating cancer patients at present is in measuring tumours. Producing accurate measurements is important in tracking whether a tumour is growing or shrinking, which indicates how it is responding to treatment, and in creating more effective treatment pathways. AI technologies are helping create new systems to measure tumours, analysing scans and images and assisting medical experts to delineate or segment tumours, and to assess how they are responding to treatments.

Interacting With Computers And Devices By Voice Recognition
Advances in AI technologies, such as deep learning, have dramatically increased the accuracy of automatic speech recognition systems that allow virtual personal assistants on smartphones and other devices to recognize and respond to commands. Steve Young (Professor of Information Engineering, University of Cambridge, and Senior Member of Technical Staff – Siri Team, Apple) is helping create these systems and explained that the technologies that underpin these conversational agents are improving quickly. In the future, there may be more sophisticated systems that can communicate with users in natural language and that have the authority to execute transactions on behalf of their owner. Unlocking this potential requires further technical advances, for example in learning human conversational skills and improving systems’ common sense reasoning. There are also broader questions about the ethics of data used by these systems – learning to answer queries requires access to large pools of data, which could raise privacy concerns – and questions about how society expects such systems to answer ‘unethical’ questions.

Supporting Advances In Science
AI technologies are already helping scientists observe and understand the universe, extracting insights from large datasets to detect events or areas of interest for further research. In analysing data from the Keppler satellite, for example, AI algorithms can help pick out small signals that indicate the existence of planets from the large amounts of noisy data picked up by the satellite. In future, such techniques may become even more important as a tool to extract insights from ever-growing datasets. The Square Kilometer Array, for example, will collect huge amounts of data from hundreds of stations, with an expected raw data flow over five times the size of 2015’s global internet traffic. Researchers – such as Suzanne Aigrain (Professor of Astrophysics, University of Oxford), who is applying AI to astronomy – will rely on algorithmic systems to process this data, identifying which data to keep for further analysis, by screening for known phenomena or important and unexpected events. To make best use of these techniques, scientists will need to understand how to effectively implement AI systems in their work. This raises questions about how best to combine subject-specific expertise with technical AI know-how, and how to design systems that produce results, which can be interpreted by their users.