Autonomous vehicles will require the trust and confidence of governments and society to support broad and wide spread adoption. Aligning development with acceptable values will play a vital role in the long term success of the AV sector.
The interface between regulation and ethics has taken a centre stage role in debates involving AI. Regulation itself does not operate objectively outside of normal discourse, rather it should seek to reflect and codify the values of society. As in the debates around the use of AI in others areas, whether that be criminal justice30 or healthcare31, the ‘tolerance’ built into these systems needs to reflect society’s views of acceptability, and these in turn inform the regulatory frameworks.
In the race to be the global leader in AI, initially the majority of governments have been hesitant to regulate, for fear of stifling development and growth. With some estimating that the sector is valued at $15.7 trillion by 2030 it is a big stakes game.
Source: 2018 AI predictions , PWC © 2018 PwC. All rights reserved
Increasingly however, there is an appreciation that some norms and standards are required, although at this stage few have been hardwired into black letter law. Recent scandals over the use of personal and social data by Facebook and Cambridge Analytica has brought ethical considerations to the fore. This sort of interest is not likely to dwindle, as AI applications become more integrated – in areas such as driving and transport systems - public scrutiny is likely to increase.
The debates around transparency, accountability, explainability, data ownership and governance, cyber security, privacy and bias are increasingly demanding a responsible and open approach to AI development and implementation by industry. Just as in other areas there will be a range of trade offs between competing values. All algorithms are inherently value laden in some way32. From the design stage onwards certain parameters are used, others are excluded and some favour one set of interests over the others. Sometimes these are explicit choices, other times, implicit or hidden in either assumptions or buried in the data. These sorts of complexities need to be fully understood during both the design and development stage and in the policy responses which mirror and guide the technology.
Multiple attempts at the national33, regional34 and global level35 are being made to establish some form of baseline – in the main opting for high level principles. In addition industry itself is seeking to define what ‘ethical’ looks like in an attempt to navigate decisions it has so far struggled with. From establishing ethics boards, with varying degrees of success, industry wide collaborations such as the Partnership for AI, in house academic research such as Google-owned DeepMind’s Ethics and Society division or company specific frameworks such as Microsoft’s 6 principles of trustworthy AI – there is a plethora of approaches. In which ever approach is adopted, deliberative policy making methods, designed to engage end users and the broader citizenry will play an important role in leading to properly informed decisions and validating the resulting standards.
Trade off’s between data privacy and efficiency, or accuracy and speed, or safety and time to market ought to reflect the values of the society in which the technology – in this case AV - is deployed. The establishment of an appropriate ethics framework is essential to commanding the trust and confidence of users, and in turn the success of the industry. In the case of AV’s / CAV’s this plays a vital role in demonstrating to a sceptical consumer base36 that the technology will be safe to use. Without engaging and calibrating for what is deemed ‘acceptable’ the industry risks losing its democratic legitimacy, and falling foul of the pitfalls encountered by GM foods in the UK.
There is a slow building consensus in places such as Europe that ‘good AI’ underpinned by a robust ethics framework can be pro growth and provide a competitive edge. In reality, it is clear that China has a huge advantage in terms of spending power and data utilisation. For those with smaller purses the competitive edge will have to come from somewhere else, and well designed ethics and regulatory frameworks might offer that opportunity.
In the short term the outlook remains patchy, but there are emerging degrees of consensus which can be built into industry wide standards and ethics – either driven by governments, industry or together in partnership.