AI systems that demonstrate significant bias or lower than claimed accuracy, and resulting in individual and societal harms, continue to be reported. Such reports beg the question as to why such systems continue to be funded, developed and deployed despite the many published ethical AI principles. This paper focusses on the funding processes for AI research grants which we have identified as a gap in the current range of ethical AI solutions such as AI procurement guidelines, AI impact assessments and (...) AI audit frameworks. We highlight the responsibilities of funding bodies to ensure investment is channelled towards trustworthy and safe AI systems and provides case studies as to how other ethical funding principles are managed. We offer a first sight of two proposals for funding bodies to consider regarding procedures they can employ. The first proposal is for the inclusion of a Trustworthy AI Statement’ section in the grant application form and offers an example of the associated guidance. The second proposal outlines the wider management requirements of a funding body for the ethical review and monitoring of funded projects to ensure adherence to the proposed ethical strategies in the applicants Trustworthy AI Statement. The anticipated outcome for such proposals being employed would be to create a ‘stop and think’ section during the project planning and application procedure requiring applicants to implement the methods for the ethically aligned design of AI. In essence it asks funders to send the message “if you want the money, then build trustworthy AI!”. (shrink)

In the last decade, the UK Government has attempted to implement improved processes and procedures in modelling and analysis in response to the Laidlaw report of 2012 and the Macpherson review of 2013. The Laidlaw report was commissioned after failings during the Intercity West Coast Rail (ICWC) Franchise procurement exercise by the Department for Transport (DfT) that led to a legal challenge of the analytical models used within the exercise. The Macpherson review looked into the quality assurance of Government analytical (...) models in the context of the experience with the Intercity West Coast franchise competition. This paper examines what progress has been made in the 8 years since the Laidlaw report in model building and best practise in government and proposes several recommendations for ways forward. This paper also discusses the Lords Science and Technology Committees of June 2020 that analysed the failings in the modelling of COVID. Despite going on to influence policy, many of the same issues raised within the Laidlaw and Macpherson Reports were also present in the Lords Science and Technology Committee enquiry. We examine the technical and organisational challenges to progress in this area and make recommendations for a way forward. (shrink)

Call for Written evidence - Risk Assessment and Risk Planning.

To date researchers planning experiments have always lived by the mantra that 'using equal sample sizes gives the best results' and although unequal groups are also used in experimentation, it is not the preferred method of many and indeed actively discouraged in literature. However, during live study planning there are other considerations that we must take into account such as availability of study participants, statistical power and, indeed, the cost of the study. These can all make allocating equal sample sizes (...) difficult, and sometimes near impossible. This, some might say, means that the study would not adhere to rigorous statistical standard (Rosenbaum and Rubin, 1985). However, here we present evidence that, not only is this a false assumption, but that we may actually gain more power in the study by actually using unequal groups. Here, data from a Sepsis Biomarker study is used, in which the aim is to predict, by biomarker level and presence, whether the patient would go on to develop sepsis. It was found that larger control groups may give more power to studies looking for an effect in the mid range but not for large or small effects. This study shows merit in the hypothesis that more power can be achieved when a larger control group is used. Published by the Ministry of Defence at dstl - Reference DSTLTR92592 P2PP2R-2016-02-23T13:39:45. (shrink)

Marie Oldfield, Pro Bono Economics & Refugee Council. Over half the total applications for asylum the UK receives each year are initially rejected, yet nearly a third of these initial rejections are subsequently overturned on appeal. This process that fails to get decisions right first time imposes significant costs, not just on the applicants themselves, but also more widely on UK taxpayers. Asylum seekers are not entitled to welfare benefits nor employment except in some limited cases, and are often placed (...) in unsuitable conditions with lack of appropriate support to prepare their case for asylum. These unfortunate circumstances mean that they are unable to gather and present all the required evidence first time, even when the case for asylum is legitimate. As a result, the application is rejected and thus begins a cycle of appeals that might eventually lead to a successful outcome for the asylum seeker but will, in the process, burden the taxpayer with unnecessary administrative cost and cause the asylum seeker needless distress and prevent them from accessing suitable economic opportunities. • Our study estimates that there are approximately 4,000 initial rejections that are successfully appealed each year and, with each appeal incurring direct administrative costs of around £1,000, the price tag of incorrect initial decisions adds up to a hefty £4 million per year in unnecessary administrative costs alone. • There are other costs, too, that negatively impact asylum seekers and taxpayers, albeit in a more indirect manner. The longer the appeals process is dragged out, the more delay there is to successful asylum seekers entering the labour market which means lost economic activity and forgone tax revenue for the Exchequer. Long periods of unemployment also result in skills attrition and lower the probability of the asylum seeker successfully entering the labour market in the future. • A lengthy appeals process heightens stress and worsens mental health in asylum seekers, a group arguably already more vulnerable on this front. Unnecessary human suffering aside, there is also a monetary cost borne by the taxpayer should NHS mental health services become involved. Thus, there is a pressing and viable case for providing asylum seekers a stable environment in which to complete their applications for asylum. This would increase the likelihood of reaching correct decisions in the first instance, avoid unnecessary administrative costs of an arduous appeals process, enable successful asylum seekers to access the labour market and contribute to the Exchequer, and avoid placing extra demands on already strained NHS mental health services. (shrink)

With an increased emphasis on upstream activity and Defence Engagement, it has become increasingly more important for the UK Ministry of Defence (MOD) and government to understand the relationship between conflict and regional instability. As part of this process, the Historical and Operational Data Analysis Team (HODA) in Defence Science and Technology Laboratory (Dstl) was tasked to look at factors that influenced the regional spread of internal conflicts to help aid the decision making of government. Conflict contagion is the process (...) by which a conflict in one state (State A) influences the outbreak of conflict in another state (State B). The aim of looking at these factors was to produce a tool to assist in the cost/benefit analysis of intervening in future conflicts. The task was carried out by quantitatively studying a selection of contagion and non-contagion case studies over a set of 14 numerical variables that covered structural, political, socioeconomic and cultural factors of both States A and B. All case studies took place after the end of the Cold War. The dataset used contained both binary and continuous data. A range of statistical techniques were used including correlation, regression and principal component analyses along with graphical analytical methods. Unfortunately creating a tool to predict contagion has proven difficult and so the scope of the study was changed-as no statistically relevant link was found. However, it was shown that some links do exists between the variables and the existence of contagion, and it also appears it is a combination of variables which has an effect on the likelihood of contagion. Conflict contagion is a rare event and was only recorded in less than 1% of cases present in the dataset. Of the variables analysed, the most significant factor was the number of refugees present in State B who originated in State A for the year before the case study. It is recommended that further work be conducted. It is possible that a more detailed investigation, using a broader range of techniques and/or a wider selection of variables might be successful in assessing the likelihood of conflict contagion. (shrink)

Historically, Anthropomorphism is a technique used by humans to make sense of their surroundings. More recently, anthropomorphism has been widely used as a technique to influence consumers to purchase goods or services. These techniques can entice consumers into buying something to fulfil a gap or desire in their life, ranging from loneliness to the desire to be exclusive. By manipulating belief systems, consumer behaviour can be exploited. Anthropomorphism can encourage society to use inappropriate emotional attachments and faulty frameworks to understand (...) complex concepts, which in turn reduces risk awareness and understanding of how that technology might impact the individual or society. This technique also separates and divides individuals and society; it plays into a deep belief system that all humans wish to be distinct and yet be part of a group. The consequences of this are found in the field of dehumanisation, which is extremely damaging for society and individuals alike. This research has found potential serious consequences for society and humanity which indicate an urgent need for further research in this area. (shrink)

Complex systems are difficult to manage, operate and maintain. This is why we see teams of highly specialised engineers in industries such as aerospace, nuclear and subsurface. Condition based monitoring is also employed to maximise the efficiency of extensive maintenance programmes instead of using periodic maintenance. A level of automation is often required in such complex engineering platforms in order to effectively and safely manage them. Advances in Artificial Intelligence related technologies have offered greater levels of automation but this potentially (...) pivots the weight of decision making away from the operator to the machine. Implementing AI or complex algorithms into a platform can mean that the Operators’ control over the system is diminished or removed altogether. For example, in the Boeing 737 Air Max Disaster, AI had been added to a platform and removed the operators’ control of the system. This meant that the operator could not then move outside the extremely reserved, algorithm defined, “envelope” of operation. This paper analyses the challenges of AI driven condition based monitoring where there is a potential to see similar consequences to those seen in control engineering. As the future of society becomes more about algorithm driven technology, it is prudent to ask, not only whether we should implement AI into complex systems, but how this can be achieved ethically and safely in order to reduce risk to life. (shrink)

Government transparency and accountability during Covid 19: The data underpinning decisions.

Artificial Intelligence is becoming widespread and as we continue ask ‘can we implement this’ we neglect to ask ‘should we implement this’. There are various frameworks and conceptual journeys one should take to ensure a robust AI product; context is one of the vital parts of this. AI is now expected to make decisions, from deciding who gets a credit card to cancer diagnosis. These decisions affect most, if not all, of society. As developers if we do not understand or (...) use fundamental modelling principles then we can cause real harm to society. Recently more serious effects of AI have been observed. Dehumanisation is the human reaction to overused anthropomorphism and lack of social contact caused by excessive interaction with, or addiction to, technology. This can cause humans to devalue technology and to devalue other humans. This is a contradiction of the use of ‘social robots’ and ‘chatbots’, indicating that the negative effects of this technology would outweigh any perceived positive effects. Also, within cyberspace, anthropomorphism and similar techniques based on deep philosophical principles can, and are, being used to alter the behaviour of humans. These techniques are used to manipulate human behaviours at a basic level in the human mind. As these types of techniques are becoming more widespread, it is clear that we are entering unchartered territory that holds a vast array of consequences for society. (shrink)

Why interdisciplinary research in AI is so important, according to Jurassic Park. -/- “Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should.” -/- I think this quote resonates with us now more than ever, especially in the world of technological development. The writers of Jurassic Park were years ahead of their time with this powerful quote. -/- As we build new technology, and we push on to see what can actually (...) be achieved there is an undertone of sales to whatever we build. The end product must be sold somewhere and to someone. This can derail any good intentions, just as building a resort full of dinosaurs was sold as a fun attraction, we see later in the film that it became a resort of terror. -/- In the field of AI we are certainly late to the party with ethics and regulation. Indeed, even existing modelling protocols have been, in many cases, circumvented or potentially ignored. This has led to a widening of the gap in the interdisciplinary field of AI. This is compounded by pop culture representations of AI and the Ethicist’s potential lack of knowledge surrounding technical progression in the field of AI. There are now seemingly two separate branches that ought to be in sync. The first branch is a group of Philosophers led by the Author of Superintelligence, Nick Bostrom, who believe that there is a singularity where AI takes over the world and starts to kill off humans. The second branch is the Technical Cohort who are led by companies such as Google and Deepmind. The remit of these developers is to see what can be developed and produced. Ultimately a separate sales team will determine what products can be sold. -/- We have seen multiple failures of AI due to lack of interdisciplinary discourse. Some cases include misallocated or cut off financial support and healthcare. This indicates the prevalence of issues that can grow and become ever more complex. -/- So, we have to ask ourselves two questions: One, do the Philosophers have a point? Two, does that point refer to humans programming AI in such a way that the destruction of the human race is its optimal aim in order to create a better future world? -/- Is the development of AI that different to an island of genetically engineered Dinosaurs? -/- The only solution, in my view, is interdisciplinary discourse. (shrink)

Over the past few years we have seen an increasing number of legal proceedings related to inappropriately implemented technology. At the same time career paths have diverged from the foundation of statistics out to Data Scientist, Machine Learning and AI. All of these new branches being fundamentally branches of statistics and mathematics. This has meant that formal training has struggled to keep up with what is required in the plethora of new roles. Mathematics as a taught subject is still based (...) in decades old teaching specifications and has not been updated centrally as a curriculum to include new technologies, coding or ethics. This subject area is firmly split between ICT and Mathematics in secondary school, continuing on to be split between Computer Science and Mathematics at University. As we move forwards with technology we see these once seperate fields becoming increasingly intertwined. We propose that existing provision for concepts such as ethics and societal responsibility in analysis currently exist but have not been incorporated into the mainstream curriculum of School or University. This is partially due to the split between fields in an educational setting but also the speed with which education is able to keep up with Industry and its requirements. Principles and frameworks of socially responsible modelling beginning at school level means that ethics and real-life modelling are introduced much earlier than normal. Integrating these concepts with philosophical principles of society and politics ensures a suitable background for future modellers and users of technology to draw on. Modelling is currently undertaken in technical sciences at University but the Subject Benchmark Statements are not current (Subject Benchmark Statements describe the nature of study and the academic standards expected of graduates in specific subject areas. They show what graduates might reasonably be expected to know, do and understand at the end of their studies). Even in 2019 the UK did not yet have Benchmark Statements that discuss the learning to be done in Higher Education around AI and advanced Machine Learning. Where there is provision for AI and Data Science within degree courses ethics is not generally highlighted as a key concept. As such there can be a lack of focus on the teaching of modelling or ethics as specific skills. The skills required to use a basic statistical model, for example would not be sufficient to start from scratch and build an ethical model reflecting real-world scenarios with which to inform policy or organisational decision making. This is a skill in itself and includes such aspects as awareness of data quality, ethics, user implementation problems, context and an understanding of the environment that the model is being created in. As the field of analytics has progressed so quickly in the last decade modules such as those covering AI have simply been bolted on to Maths or Computing degrees rather than being fully detailed as key areas of study or driving new, more integrated courses of study. This paper posits that gaps at primary, secondary and tertiary educational levels need to be addressed. Implementing and integrating key concepts from school level is essential so that areas such as assumptions, caveats, quality assurance and answering the right questions with constructive challenge become a cultural fixture. This not only helps developers of technology but users of rapidly developing technology also. In addition, leadership and soft skills as part of this education will ensure that a cultural shift can take place and promote continuous improvement in analysis within organisations. The addition of key concepts throughout the educational system and the updating of potentially outdated curriculums is key to ensuring a functional society. A society where every citizen is a user of tech and those that develop it can be ethical and socially responsible in its development. (shrink)

Call for evidence Written evidence - Data Transparency and Accountability: Covid 19 -/- .

This paper discusses Claytons theory on Comprehensive enrolment of children by their parents. This paper supports Claytons view that we should not enrol children. However, Cameron raises objections which cause problems for the application of this framework. Namely, the cost of giving up a belief, choices made for us in childhood and the application of the PRR (Public Reason Restriction) to the way the parent-child relationship should function. Some modifications to Clayton’s framework and further debate is required to fully address (...) these issues. The conclusion is that we should be able to enrol children in activities that would be of low future cost if rejected but we should not enrol children in activities of high future rejection cost. This enrolment is tempered by the statement “the fundamental motivation of parents should be to conform with public reason i.e. to treat their children in accordance with norms that are capable of acceptance by any free and equal person”. As Clayton states: “I am not ruling out the imposition of a comprehensive doctrine on the child. I am rejecting its imposition in the absence of an argument from public reason”. The structure of this essay is as follows: In Section One I explore Clayton’s theories of end state autonomy and autonomy as a precondition, I then look at the plausible relationship between the state-citizen and child-parent relationships. The Public Reason Restriction is then examined in connection with comprehensive enrolment. In the next section, I look at objections to Clayton’s view from Cameron and any subsequent replies to this from Clayton. I then conclude by discussing the differences between the two views and add my own view to this. (shrink)

What is coercion and why do we care? Coercion is widespread and used especially when raising children, but on its darker side coercion can have devastating consequences. We are worried about coercion as it can invalidate consent. This is seen in the USA where campus rape cases have soared in recent years and brought consent and coercion back to the forefront of debate. Coercion is a hotly debated legal, political and ethical concept. However, in all this debate we have seen (...) little in the way of what coercion means exactly and how we determine it is taking place. Wertheimer and Garnett have interesting views about how coercion can be understood, from the morally wrong side of coercion to the situations where people are left much worse off because of its use. Only the morally wrong situations are subject to the law but there is a whole set of coercive situations that could take place which look bad to the outsider but are not necessarily morally wrong. This is the area of coercion I wish to discuss further. Wertheimer states that there are two types of coercion, moralised and non moralised. Moralised coercion concerns whether it was ‘wrongful’ to propose the offer and the non-moralised notion examines whether the coerced had any other reasonable choice. In a similar vein to Wertheimer, Garnett believes that coercion has two roles, Deontic (wrong) and Eudaimonic (bad). This would vaguely map to Deontic as Moralised and Eudaimonic as Non Moralised. Garnett discusses the eudaimonic version at length believing it is relatively unexplored in situations of coercion. This paper consists of 4 sections: • Section 1 discusses what Coercion is • Section 2 reviews Wertheimer’s view of Moralised and Non Moralised Consent • Section 3 discusses Garnett’s View • Section 4 considers a cost-benefit analysis of eudaimonically wrong situations. In conclusion it is apparent that, in order to have a full picture of what kind of situations coercion actually covers we must not simply rely on the deontic concept. The Deontic way of looking at situations will tell us what is morally wrongful and possibly redressable in court but many situations simply sit badly with us. It is here, in this arena, that the eudaimonic concept comes into its own. (shrink)

Emotions are an everyday occurrence. Much work has been done into what the point of emotion is and what part emotions might play in our lives. The great impact emotions have on our lives means it’s not surprising that great philosophers have studied them over the centuries. Anger is an emotion that we encounter every day and most of us are very familiar with. Anger is a response to some ‘wrongfully’ inflicted damage to someone or something that one cares about. (...) When this happens, it is natural for us to feel the need for ‘payback. It is generally agreed by most philosophers that this wish for retribution is a key part of feeling anger. If there is no wish for retribution in your mind when you think you experience anger then perhaps you are experiencing grief or some other emotion. Nussbaum states that anger is a central threat to decent human interactions and posits that a necessary component of anger is the wish for retribution. She states that a wish for retribution can occur independently of anger but anger cannot occur independently of a wish for retribution. In this case – we would be feeling different emotion. This paper has two sections; In section One I will discuss in detail Nussbaum’s view, the two concepts of anger she presents: ‘Payback Anger’ and ‘Status Anger’. The former being anger with a retributive element and the latter being where one would react to an insult or wrong with anger in order to ‘put someone back in their place’ so to speak. This is followed by a brief word on why Nussbaum thinks anger might be beneficial in some circumstances. Section Two discusses cases of anger without the retributive element and how Nussbaum might respond to these cases. We will see how we can be angry at something within our circle of concern but simply want an end to the issue such as a war. This may not involve retribution but still generates anger. I discuss how we can be angry at a flatmate or family member but not want retribution. “Self-Retributive anger” comes up as another case where there does not seem to be retribution involved in anger towards the self. I also discuss here what retribution might consist of and how Nussbaum’s view is unclear on this. The conclusion will summarise the objections to Nussbaum’s view and the objections to that view culminating in a request for further empirical research to determine if we can establish the nature of the feeling of anger in humans. (shrink)

Increasingly, models have been highlighted that not only disadvantage society but those whom the model was originally designed to benefit. An increasing number of legal challenges around the world illustrates this. A surge of recent work has focussed on the technical, legal or regulatory challenges but not necessarily the real-world day to day challenges for practitioners such as data collection or fairness by design. Since the publication of the Holstein et al.’s study in 2019, additional legislation, regulation and multiple bodies (...) have been created to address practitioner challenge. This study asks what, if anything, has improved for practitioners between 2019 and 2022. Study 1 conducts an investigation into real-world needs within industry and asks whether practitioners are now able to mitigate challenges in a more robust manner. A further pilot study on the perception of AI examines whether perception of AI impacts practitioner work. The results show increasing and continuing interdisciplinary issues. Where increased regulation and legislation might have seemed reasonable, the result for practitioners is indecision and overwhelm. Based on these findings, we highlight directions for future research in this area. The most problematic area being human factors. (shrink)

In this paper I will first examine Williamsons case where he posits that ‘Justified Belief’ is not acceptable to use in the cases where one is deceived in some way, regardless of how the belief has been formed. Williamson aims to eradicate the use of the term ‘justified’ in these cases and instead impose the term unjustified but blameless. I counter this by suggesting that Williamson sets the bar too high to attain knowledge and does not give enough weight to (...) the manner in which beliefs are formed. I then offer another way to describe the situation. The structure of the essay is as follows: I will present Williamson’s main argument, define his concept of ‘norms’ and present some examples to illustrate this. I will then move on to Williamson’s formulation of blamelessness whilst unjustified in a belief as an alternative to simply justified or unjustified beliefs. I will again illustrate this with examples. Then we will explore some objections to this theory. (shrink)

The dystopian tale has a special place in our shared cultural heritage. -/- Many of us will have a favourite, or perhaps several. I myself adored the 1984 and The Handmaid’s Tale books as a youngster; moved on to JG Ballard then discovered Philip K. Dick thanks to Minority Report; and in recent years was floored by Black Mirror episodes and videogames such as The Last of Us. -/- The thrill can be explained by one question: ‘What if this horror (...) was actually happening?’ -/- “People say Black Mirror and The Handmaid’s Tale are conspiracies, science fiction – but as a philosopher, I can see a lot of the elements in these films and books that are actually happening now,” Marie Oldfield, AI ethics consultant to the UK government and other organisations, tells BusinessCloud. (shrink)

Complex platforms are very difficult to manage and maintain. This iswhy we see teams of engineers, many highly specialised, that carry outthis role for industries such as aerospace, nuclear and subsurface. It is acritical undertaking to maintain the aforementioned systems, which oftenhave components at varying degrees of degradation. To maintain com-plex systems, Condition Based Monitoring (CBM) (a type of predictivemaintenance that uses sensors to measure the status of an asset over timewhile it is in operation) is most frequently used. Artificial (...) Intelligence(AI) models that have been developed in the area of CBM are currentlynot well explained, nor well understood by users or operators. When AIis brought into a complex system we observe varying degrees of success.The level of success rests on the complexity of the system, the trainingand understanding of the end operator as well as the maintenance pro-cesses around the system. Implementing AI or complex algorithms into aplatform can mean that the Operators control over the system is dimin-ished or removed altogether. For example, in the Boeing 737 Air MaxDisaster, AI had been added to a platform and removed the operators’control of the system. This meant that the operator could not then moveoutside the extremely reserved, algorithm defined, ‘envelope’ of opera-tion, leading to loss of life. Therefore, the implementation of AI leading toany removal of operator system management in complex systems, espe-cially related to aerospace and subsurface industries, has to be consideredcarefully. In this paper we analyse the risks of removing operator systemcontrol and implementing algorithms, or AI, in complex systems. (shrink)

Digital technology has advanced more rapidly than ever before in the last 20 years. However, people may not be likely to understand the risks of this technology. This is due to developers not being transparent about functionality. This development has also far outpaced educational provision in schools, leading to a world in which we are encouraged to engage with digital technology, but we don’t quite understand it, how it works or what it does; therefore, if our data is harvested and (...) sold it is much more difficult to understand when, why and how this is happening. (shrink)

Will AI take away your job? The answer is probably not. AI systems can be good predictive systems and be very good at pattern recognition. AI systems have a very repetitive approach to sets of data, which can be useful in certain circumstances. However, AI does make obvious mistakes. This is because AI does not have a sense of context. As Humans we have years of experience in the real world. We have vast amounts of contextual data stored in our (...) brains that make it possible to predict and to know boundaries of the real world so that even if we have never been in a particular situation, we are still able to deal with it. The unknown situation is where AI will fall down. In engineering, AI is being developed to monitor and control certain systems, this could extend to Nuclear Power Plants for example. If we examine control systems for Nuclear Power Plants on vessels, the system would likely be programmed using the safety of the reactor as the main priority. Automation is crucial in engineering systems like this, as the likelihood and cost of human error is high and human reaction times are slow in comparison to the system. However, there is always the possibility of unforeseen external factors that may override plant safety which cannot be programmed. If we look at a plant onboard a Submarine, for example, we see such additional factors as ship safety. In one example, the power plant might breach safety parameters, in which case an automatic system may shut down the reactor. However, there might be a greater urgency such as a flood in another compartment or an attack upon the vessel, that would override shutting down the reactor. The AI system can neither observe nor utilise information from the external environment to make this type of judgement, whereas the operator may be completely aware. To programme context into AI systems would be near impossible. We still need operators as we still need to understand the environment around us, which may contain an infinite number of possibilities. In a further example of this; if we examine the ICO (Information Commissioner's Office) guidance, it is clear that in clinical settings a human being will always be required to validate findings and to provide context. Sometimes having a lot of data just doesn't replace social interaction, intuition and experience. (shrink)