As we turn the corner into a new year, and new decade, we are finding ourselves emerging into a dynamic and changing world; a world full of many novel challenges in the field of systems safety, with no clear direction on how these are going to be resolved. There are now many “disruptors” that challenge the traditional methods of safety management: Artificial Intelligence (AI), Machine Learning (ML), autonomous systems, highly data-centric systems (DCS) and remotely piloted commercial drones, to name a few.
The world seems eager to embrace these technologies, and momentum seems to be gathering as the value of these technologies is already being realised. We are already seeing successful application of AI in a wide range of sectors such as agriculture, transport and healthcare. For example, one report suggests AI is now better at predicting some cancers than trained clinicians.
The fundamental question is therefore: how will we assure systems using these disruptive technologies? How do we prove to ourselves that a future full of such systems will be sufficiently safe, especially when it may be difficult to apply traditional methods of safety management?
A number of articles in this edition of the newsletter provide insight into the challenges of assurance in this changing world.
The first article from Shakir Laher and Mark Sujan, discusses the challenges of assuring health IT systems that use Artificial Intelligence and Machine Learning technologies in delivering patient care, such as the autonomous use of an infusion pump to deliver medication to a patient.
Dr Mike Standish and Dr Mark Hadley then discuss the challenges of developing non process-based assurance arguments using diverse evidence, to accommodate aviation regulatory bodies who are creating new paths to developing software safety assurance cases.
Alastair Faulkner then describes the assurance challenges with highly data-defined and data-driven systems in a world prevalent with ML and autonomy.
Our final article from Dave Banham, provides an introduction to an exciting area of work that is formalising the language we use to talk about risk. This could potentially allow the risks arising from system safety and security concerns to be described unambiguously using a common language, thus harmonising collaboration between the two domains.
In this edition we have SCSC event reports covering the Senior Leadership Forum and two seminars on Data Safety Evolution and Creating and Maintaining an Effective Safety Culture.
I have also introduced a new ’60 second interview’ feature, and Tim Kelly, the previous SCSC Director, has kindly agreed to be my first interviewee.
Paul Hampton SCSC Newsletter Editor