1.What is the current state of artificial intelligence and what factors have contributed to this? How is it likely to develop over the next 5, 10 and 20 years? What facto rs, technical or societal, will accelerate or hinder this development?
1.1 In order to answer this question, it is important to define and understand exactly what we are talking about when it comes to ‘Artificial Intelligence’ (“AI”). This term is being used to categorise a range of new technologies that are all inter-dependent and, in our view, is often being used incorrectly.
1.2 The Committee on Legal Affairs of the European Parliament produced a report containing recommendations to the Commission on the Civil Law rules on robotics on 27 January 2017 (“the European Parliament Report”), the House of Commons, Science and Technology Committee Report on robotics and artificial intelligence was produced in September 2016 (“the House of Commons Report”); and the RAS Robotics and Autonomous Systems Report by the Special Interest Group produced in July 2014 (the “RAS-SIG Report”) are focussed on what are known as ‘Robotic Autonomous Systems’ (“RAS”), which contain within them the separate disciplines of robotics, big data, machine learning, autonomous systems, the Internet of Things and AI. It is our view, backed up by the references in the above reports, that AI is a subset of RAS and that this distinction needs to be made because of the confusion that may follow as a result.
1.3 As it is one of the main recommendations from the European Parliament Report that certain robots should be registered at a ‘to be established’ regulatory body, then the question has to be asked, what is a robot? If a robot is to be granted a certain legal status (which is another recommendation), then what exactly is it?
1.4 AI is not the same as a robot. AI is, in its purest sense, software that allows machines (robots/autonomous cars/computers (IBM Watson, Google Deepmind)) to sift through vast amounts of data in order to establish certain correlations or data points, which in turn provide hitherto unimagined insight into what may be happening in certain areas, for example, the interrogation of the human genome for scientific research purposes.
1.5 Autonomous vehicles (driverless cars) use AI to analyse vast amounts of data collected by the car’s sensors as it drives down the road in order to recognise objects and road markings (amongst other things) so that the car can navigate around its surroundings – but to term this technology as being AI is to miss the point. AI is used within these RAS to “make decisions” based upon algorithms which have been programmed into computer systems. If the algorithms do not work properly, then that is a functionality issue and the software is at fault. The current level of ‘AI’ within a driverless vehicle does not allow a car to make its own decisions as a human being would – it is merely hardware and software working together, albeit in an extraordinarily complex way.
1.6 The current sophistication of AI is, therefore, not really that different from current technology especially when it comes to reviewing issues of liability for non-performance. If the software does not work, then the software developer is at fault because there is a functionality issue.
1.7 However, in the future, when machines start to interact with other machines without any human involvement and actually begin to perform functions based upon these interactions, then traditional legal, societal, ethical and technological frameworks become eroded. All of these have been developed to date with the human being as the central actor on this stage – companies, partnerships and other legal constructs aside.
1.8 The minute that RAS moves away from this human centric construct is the minute that things change completely. As RAS become capable of making their own decisions, enter into contracts, create software and hardware and generally act as humans traditionally have, this creates issues for traditional constructs like law and this is why the importance of definitions comes to the fore.
1.9 Regarding the factors that may accelerate or hinder this development, these may be described as being largely two sides of each particular issue. For example, public trust in the autonomous vehicle industry will either exist (and therefore help to accelerate the industry’s development) or it will not. If it exists, this will help the technology develop. If it does not exist, then future development is at risk.
1.10 Public trust is one example of this, others would include the following:
- Well informed regulation/legislation;
- Continued advances in technology (battery technology, computing, engineering etc);
- Education and creation of a skilled workforce;
- Continued investment in research and development; and
- Reduction in cost of usage.
2. Is the current level of excitement which surrounds artificial intelligence warranted?
2.1 As outlined above, AI is part of a larger, more complex and more integrated set of technologies which have been loosely defined as representing ‘Industry 4.0’.
2.2 These technologies working together, will, we believe, have a profound impact over the population of this planet and so we believe that this current level of excitement is warranted.
3. How can the general public best be prepared for more widespread use of artificial intelligence?
3.1 Increasing public awareness of what AI and RAS are and how they are likely to be used, will be helpful in managing the public’s expectations, reversing our inherent resistance to change and help to reduce our fear of the unknown.
3.2 For example, in dealing with the fear of potential, large scale job losses, it should be made clear that human development has been characterised by much technological change over the centuries and much like the automobile was seen as a threat to jobs and the ‘way of life’ at the beginning of the 20th century, such fears were shown to have been misplaced given that a new industry was created, alongside millions of jobs. The same should be true of developments in the application of AI and RAS.
3.3 The easiest way for the public to get used to and prepare for the advances in these new technologies is to see them in action, to use machines that are designed as RAS, to be educated about the benefits and not to be continually bombarded by horror stories or what may happen ‘when the robots take over’.
3.4 Much like genetic engineering, chemical engineering, atomic research and development and countless other technologies, there is always a ‘bad’ side to what can be developed but there are also the tremendous benefits to humanity that should be taken into account – and Government has a large part to play in reinforcing this and regulating against any harmful side effects.
3.5 STEM subjects taught at schools will enable more people to understand what is being talked about, as well as training the necessary workforce to develop and manufacture these machines and supporting systems.
3.6 Data collection, protection, analysis and storage are extremely important issues to be addressed and understood. Current attempts to try and create ‘trusted’ parties who will, on the one hand, gather huge amounts of data and, on the other hand, use such data to create new products and services, should be examined closely.
3.7 Intellectual property rights and commercial imperatives should continue to play a role in framing how datasets are made available to public and private entities and the value in such rights should not be eroded by developing additional regulatory bodies to oversee these areas.
4. Who in society is gaining the most from the development and use of artificial intelligence and data? Who is gaining the least? How can potential disparities be mitigated?
4.1 Society as a whole stands to benefit from the development and use of AI and RAS, provided that the technology is developed and implemented responsibly.
4.2 At the moment, it is the technologically savvy that are benefitting the most as new businesses spring up (some of which grow very quickly indeed) and existing technology companies deploy AI and RAS in their products and services. This pace of company development and concentration of wealth into the technology classes is expected to increase as RAS becomes more widely used – but the potential benefits to ordinary people should not be underestimated.
4.3 Healthcare solutions using robotic carers, smart cities, smart homes, driverless cars, space and sea exploration, reduced global emissions and many more areas for RAS application will provide wide ranging benefits – it is up to technologists and Government to plan ahead in order that these technologies can be used for good and jobs can be created in the UK on the back of them.
4.4 What could be problematic in the future is vast wealth concentrated in smaller and smaller businesses that employ fewer and fewer humans but run vast estates of RAS. The implications for taxation, employment and public services are self-evident. Therefore, debates around the creation of a ‘robot tax’ or a ‘universal basic income’ to everyone should be held to better prepare society for any potential implications in the future.
5. Should efforts be made to improve the public’s understanding of, and engagement with, artificial intelligence? If so, how?
5.1 Public trust is crucial to the successful propagation of RAS across society. Further, in order for the public to readily adopt the RAS technologies, clarity of accountability, liability and routes of recourse and redress are necessary. If something goes wrong, people generally want to lay blame at someone else’s door. As a result, systems must be put in place for when incidents occur and things go wrong, and the public should be aware of these routes of redress and compensation.
5.2 Trust is built through understanding, which is achieved by effective communication of the technology’s potentially substantial benefits for society, along with the negative effects and how they will be dealt with. Engagement with the public, in a way that demonstrates public opinion is being actively considered and incorporated with regard to its implementation, will be extremely important.
5.3 If the public believes that RAS will be both beneficial and safe, the development and up-take of the new technologies will be greater.
5.4 One way to effect this trust is by creating industry standards to guarantee what is safe, as judged by a regulatory body trusted to monitor and certify such standards against a code of practice for the use of RAS.
5.5 It has long been considered that public trust in new technologies is directly affected by the amount of regulation that is put in place and so industries such as the aviation industry are often cited as examples where robust regulation increases public trust in an otherwise inherently risky process.
5.6 Media outlets should be encouraged to view the positive aspects of RAS rather than the negative ones.
5.7 More funding needs to be made available to research bodies and universities, particularly in the light of Brexit and the extremely high percentage of new businesses that are funded by the EU.
5.8 Primary and secondary education should include computer science and RAS on the syllabus as well as the STEM subjects – and more emphasis should be put on the participation of woman in these areas.
5.9 More than anything, a coordinated Government approach is required across all Departments. These new technologies are currently recognised as being of particular importance to the UK economy for the future and they should not be forgotten or hampered by Brexit or a disparate, uncoordinated approach across Government.
6. What are the key sectors that stand to benefit from the development and use of artificial intelligence? Which sectors do not?
6.1 It is our view that virtually all industry sectors will benefit from RAS, much like existing technology and the use of the Internet has changed the way we live our lives across all areas.
7. How can the data-based monopolies of some large corporations, and the ‘winner-takes-all’ economies associated with them, be addressed? How can data be managed and safeguarded to ensure it contributes to the public good and a well-functioning economy?
7.1 Data is currently the steam that drives the pistons of technological evolution in our society and therefore its importance to future technological innovation cannot be underestimated.
7.2 Collecting ‘clean’ and untarnished data is expensive and time consuming. Some companies profit almost solely from such processing, which means such companies will wish to protect their industrial know-how and trade secrets and will be reluctant to make it available, free of charge, for general use.
7.3 Despite the ideal of a global community where RAS are free to use information and interact across jurisdictions free of charge, it seems more likely that a rights-driven, hard-nosed and economically focused commercial approach will occur.
7.4 Perhaps it is time to look at ways in which data can be made available outside of such approaches and so perhaps the models of usage applied to existing technology, like open source software, should be examined. In relation to data held in the public realm, it should be understood that such data is potentially very valuable, and any public – private commercial arrangements to exploit such data should be designed to extract as much value for the public as possible.
8. What are the ethical implications of the development and use of artificial intelligence? How can any negative implications be resolved?
8.1 The key ethical issues that arise from the ever-increasing pervasiveness of AI and RAS in society need to be addressed, particularly in relation to: i) safety and control; ii) privacy and consent; and iii) diversity.
8.2 Safety and Control – it will be vital for public trust that RAS are safe and can be controlled if required. There are many ethical, moral and legal issues if RAS are not inherently safe or controllable.
8.3 Privacy and consent – RAS may be used in homes and in public spaces and therefore people will not necessarily know that they are present and operating. The collection, storage and analysis of data in these circumstances needs to be highly regulated if public trust in RAS is to be established. We will interact with RAS in ways hitherto not envisaged and the anthropomorphism of robots will lead to people looking at machines in a completely different way than is the case at present. All of which needs to be taken into account when looking at laws that deal with data collection, storage and analysis.
8.4 Diversity – RAS runs on data and data needs to cover all aspects of our society, not just a rich, white, middle aged and male section of it. Decisions will be made by RAS based on such data and so careful analysis of the datasets being used will be necessary.
8.5 What is it to be human? – Such questions will need to be addressed once the use of brain-computer interface technologies, exoskeletons, robotics limbs and organs become prevalent. If people become more machine than human, does this impact upon their rights in society?
8.6 Ethics in general – will play a central role in framing the difficult questions and how society needs to deal with them. The outcomes of such debates will heavily influence the law makers and regulators of the future.
9. In what situations is a relative lack of transparency in artificial intelligence systems (so-called ‘black boxing’) acceptable? When should it not be permissible?
9.1 In order to maintain security and privacy, the ‘black boxing’ of AI may be necessary – but only to the general public.
9.2 In order to maintain public trust and in order to prove that AI and RAS are inherently safe and controllable, transparency will always be required to some degree via specific laws and regulations.
9.3 The new technologies may be used for bad as well as good and so certain aspects of it may need to be made subject to the same sort of regulatory framework as is currently applied to nuclear proliferation and other weapons grade technologies available today.
10. What role should the Government take in the development and use of artificial intelligence in the United Kingdom? Should artificial intelligence be regulated? If so, how?
10.1 Government should look at regulation specifically as it relates to each application of RAS, across industry sectors.
10.2 For example, regulation to enable the testing and development of autonomous vehicles will be different to that required for the use of care robots in a person’s home.
10.3 A ‘robot law’ seeking to regulate RAS across the board, as was envisaged by the European Parliament Report, is not necessary in the UK, we believe, because the UK has a common law system. This system has proved to be very successful in the past in applying existing common law principles to new and emerging technologies.
10.4 A nascent industry does not need more product liability laws. These may kill off development and since we already have such well-developed concepts like ‘negligence’ and ‘duty of care’ at common law, these add a flexibility of approach that will benefit the future development of new technologies.
10.5 When the time arises that RAS communicate with each other and start making decisions, entering into agreements or creating works, then existing laws may need to be adapted, for example in relation to Contract Law and Intellectual Property Law. At present, both these areas of law require the presence of humans to either form contracts or create and register patents, for example. As the technologies develop, we would suggest that specialist technology lawyers need to be consulted to advise on the best ways to approach the regulation of the same.
11. What lessons can be learnt from other countries or international organisations (e.g. the European Union, the World Economic Forum) in their policy approach to artificial intelligence
11.1 In Europe, the new General Data Protection Regulation will provide a new framework for the collection and storage of data. How such a regulation has been developed to cater for changes in technology is a good example of how difficult it is for law to keep pace with technology and lessons should be learnt.
11.2 The European Parliament Report provides a series of recommendations to the European Commission. This approach was largely adopted by the House of Commons Report and is a very helpful start in highlighting the many issues that RAS creates and how they may be dealt with.
11.3 The USA continues to be a leader in the field of RAS and it has many commentators who write extensively about the topic. Their views and experiences should be harnessed in order to provide insights into the industry.
11.4 In Europe, aside from the work being done by the European Parliament, the European Robotics Forum is an excellent example of an organisation that represents the interest of industrialists and academics operating in the RAS industry and its views should be actively sought in developing legal frameworks and approaches.
11.5 Finally, from a legal perspective, the IBA (the International Bar Association) which has a dedicated Technology Law Committee, should be canvassed for its views on the international implications of RAS and AI.