Opportunities and limitations of machine learning and other AI techniques in complex decision-making

25 Nov 2019 9:30–16:30

The Analysis Under Uncertainty for Decision-Makers (AU4DM) Network hosted its 2nd annual Workshop on 25th November 2019. This one-day workshop featured presentations and discussions from experts in the field and offered the opportunity for participants to be part of a break-out group discussion. The workshop ran three simulations in parallel, each with the same scenario but different points of view. The key objective of these simulations was: To identify what issues, events, behaviours, responses, etc. the participants perceive as being uncertain in the scenario and how they are accounted across groups.

Each delegate received a copy of both the recently updated Visualisations Catalogue and the Decision Tools Catalogue.


Monday 25 November

09:30 Arrival and registration
10:00 Welcome
10:05 David Hartley (University of Warwick and GSK)
 Analytics, Machine Learning: what can they do and what can’t they do?
10:40 Chris Dent (University of Edinburgh and Alan Turing Institute)
 Energy systems modelling: models and the real world.
11:15 Discussion
11:35 Introduction to break-out sections
11:40 Coffee break
12:00 Break-outs for scenario-focused discussion
13:15 Lunch
14:00 Break-outs
15:15 Tea
15:35 Launch of new visualisation and decision tools catalogues
15:50 Report back from break-out groups and discussion
16:20 Report on current AU4DM activities and plans
16:30 Close and departure


10:05 – 10:40 David Hartley (University of Warwick and GSK)
Analytics, Machine Learning: what can they do and what can’t they do?

Machine Learning methods are great tools to enhance analytical capabilities for decision making, however, they are not panaceas for the issues analysts must face. This talk will discuss some of the advantages and limitations of applying Machine Learning to problem solving in commercial settings from the perspective of both the decision maker and the analyst. Some common pitfalls that decision makers and practitioners fall into whilst implementing analytical/ML initiatives are discussed.

10:40 – 11:15 Chris Dent (University of Edinburgh and Alan Turing Institute)
Energy systems modelling: models and the real world

Computer modelling is widely used for planning and decision support in energy systems planning and policy, and in many other situations of decision making in complex circumstances. This project will review key issues in linking models and the real word, with the help of quotations from a number of prominent experts. While specific examples are from energy systems, the issues discussed are of general relevance. The work will also briefly discuss current work on uncertainty in government modelling at the Alan Turing Institute.


Warwick in London Rooms
The Stanley Building
7 Pancras Square
Kings Cross, London N1C 4AG