Blackett 2018: Sir Alan Wilson on OR in the age of AI


The 2018 Blackett Memorial Lecture was given by Sir Alan Wilson at the Royal Society in London. Sir Alan is Director of Special Projects at the Alan Turing Institute, the Executive Chair of the Ada Lovelace Institute and Chair of the Home Office Science Advisory Council. He is a fellow of both the British Academy and of the Royal Society. He was knighted in 2001 for services to higher education and he is also an honorary member of The OR Society.

Sir Alan began his lecture, ’OR in the age of AI’, by saying despite being a mathematician and geographer he had always considered himself as an operational researcher, so he thought it appropriate to find a text for the talk made by Patrick Blackett himself. It was “the most important qualification [of an operational research worker] is the ability to take a broad view of a problem, so that important factors will not be missed.” (P.M.S Blackett, Operations Research, 1950).

OR in the age of AI

Sir Alan Wilson FRS FAcSS FBA

Sir Alan said his presentation would deal with four topics:

  1. OR in a new age: the starting point
  2. The impact of AI
  3. OR in the age of AI
  4. The future

Improvements in computing power and the use of big data applications have helped create a recognition of the value of data and AI itself. He believed that AI was going to expand the OR toolkit because it could assist with policy decisions, design and analysis. Combining the insights of analytics and AI with imaginative thinking about design could help scientists find solutions to many problems.

If you have a big challenge, or you have a plan, analytics will not solve it. What solves it, if you can solve it, is actually combining the insights from analytics with imaginative thinking about design. In the big planning problems, we very rarely find all three – policy, design and analysis – in the same room.

2018 Blackett Lecture.jpg

Sir Alan found the retail sector interesting because you can actually see where changes are taking place. After centuries of selling products in one way, all has changed in just a five-year period with Amazon’s online retail transition and other companies moving their retail enterprise online. Public sector services, too, are ripe for change, though this has not yet happened.

AI-enabled machines can read, write and translate. AI can drive robotics which enable automated processes. AI can help us understand challenges in marketing and provide new mechanisms whereby customers can be influenced to buy, thus AI can drive sales and improve profitability.

Additionally, the new tools of analytics and AI can be used to essentially re-engineer many processes. If you add the new tools to the classic OR tools, there are many new possibilities for OR in this age of AI. We have new data everywhere; we have sensors that can work with real time data; and that leads us into a world of real-time OR. It used to take months to calibrate a model; now we are in sight of being able to calibrate models in real time.

Machine learning (ML) is making huge impacts in the use of statistical information. There are huge amounts of statistics in AI and ML applications, and the algorithms used in them are brilliant at clustering, something that is critical in terms of contextualising health data for example.

AI and ML are transformative. They feed back into all sectors, in big data, government, third sector and public sector services. However, AI and ML are not the perfect tools that can deal with everything. They are, for example, poor at dealing with ethical challenges, poor at ensuring constant data acquisition and they have no notion of privacy. There is a danger that their application in decision making can be “unfair” and “opaque”; they can introduce bias into decision making situations.

Too much AI research has been, to date, human imitative. Such imitative characteristics are not useful for intelligence augmentation or for building intelligent infrastructures. Much of current OR practice will continue, but with updated technology, deploying the basics of AI. There will be opportunities to engage with large, complex systems and their challenges, and for OR to ’think big’ as a result of updated technologies.