Are Robots Sociopaths? A Neuroscientific Approach to the Alignment Problem

Artificial intelligence (AI) is expanding into every niche of human life, organizing our activity, expanding our agency and interacting with us to an exponentially increasing extent. At the same time, AI’s efficiency, complexity and refinement are growing at an accelerating speed. An expanding, ubiquitous intelligence that does not have a means to care about us poses a species-level risk. Justifiably, there is a growing concern with the immediate problem of how to engineer an AI that is aligned with human interests. Computational approaches to the alignment problem currently focus on engineering AI systems to (i) parameterize human values such as harm and flourishing, and (ii) avoid overly drastic solutions, even if these are seemingly optimal. In parallel, ongoing work in applied AI (caregiving, consumer care) is concerned with developing artificial empathy, teaching AI’s to decode human feelings and behavior, and evince appropriate emotional responses.We propose that in the absence of affective empathy (which allows us to share in the states of others), existing approaches to artificial empathy may fail to reliably produce the pro-social, caring component of empathy, potentially resulting in increasingly cognitively complex sociopaths. We adopt the colloquial usage of the term “sociopath” to signify an intelligence possessing cognitive empathy (i.e., the ability to decode, infer, and model the mental and affective states of others), but crucially lacking pro-social, empathic concern arising from shared affect and embodiment. It is widely acknowledged that aversion to causing harm is foundational to the formation of empathy and moral behavior. However, harm aversion is itself predicated on the experience of harm, within the context of the preservation of physical integrity. Following from this, we argue that a “top-down” rule-based approach to achieving caring AI may be inherently unable to anticipate and adapt to the inevitable novel moral/logistical dilemmas faced by an expanding AI. Crucially, it may be more effective to coax caring to emerge from the bottom up, baked into an embodied, vulnerable artificial intelligence with an incentive to preserve its physical integrity. This may be achieved via iterative optimization within a series of tailored environments with incentives and contingencies inspired by the development of empathic concern in humans. Here we attempt an outline of what these training steps might look like. We speculate that work of this kind may allow for AI that surpasses empathic fatigue and the idiosyncrasies, biases, and computational limits that restrict human empathy. While for us, “a single death is a tragedy, a million deaths are a statistic”, the scaleable complexity of AI may allow it to deal proportionately with complex, large-scale ethical dilemmas. Hopefully, by addressing this problem seriously in the early stages of AI’s integration with society, we may one day be accompanied by AI that plans and behaves with a deeply ingrained weight placed on the welfare of others, coupled with the cognitive complexity necessary to understand and solve extraordinary problems.

Introduction: Black Swans and Sinatra Inferences

Public mismanagement as a threat to life and limb is a rare and highly improbable phenomenon—the proverbial Black Swan. Bridges and buildings collapse, claiming the lives of people who had every reason to believe that governmental agencies protect their physical integrity through public oversight and maintenance. Properly analyzed, however, these unlikely events reveal causal mechanisms of a general nature, strong enough to trigger fateful mismanagement even under the restrictive conditions of professional bureaucracies and democratic government. Hence the “Sinatra Inference”: When a mechanism is powerful enough ‘to make it there’—i.e., where causal leverage is supposedly low—it is likely to ‘make it everywhere’ as soon as leverage is enlarged by weaker accountability structures, lower professional standards and lesser values than human safety.

Smuggled Migrants as Victims?

After outlining the UN Protocol’s general approach to migrant smuggling, this article raises the question of whether and to what extent smuggled migrants can be said to be victims of this crime. The author argues that an affirmative answer is possible in at least three different respects: smuggled migrants can be victimized by states fighting against migrant smuggling and irregular immigration (secondary victimization); but, of course, they can also be victimized by smugglers (primary victimization), in two ways: first, if smuggling is so performed as to put their lives, physical integrity or dignity at risk; secondly, smugglers also victimize migrants by profiting of their vulnerable condition, and their need to enter a foreign country, in order to gain economic benefit.

Measuring Whistleblowing Perceptions among the Civil Service of the Republic of Kosovo

In 2018, the Parliament of the Republic of Kosovo approved the Law on the Protection of Whistleblowers, setting up the foundations of the whistleblower protection system for the public and private sectors in the country. In line with the international principles for drafting legislation for the protection of whistleblowers, the law provides three channels for reporting wrongdoing and grants protection against any form of retaliation for whistleblowers. Noting the absence of institutional data on whistleblowing in the public sector, for this research article, a survey was implemented with individual members of civil service in Kosovo (n=400), during the period from September to November 2019, to collect primary data related to factors incentivizing and / or discouraging the decision to whistleblow. Data were collected at the national and local levels of state administration, as per the scope of the definition of the civil service by Kosovo legislation. In this contribution, research results reveal that the protection against any form of retaliation guaranteed by the law is not sufficient for members of civil service in Kosovo to support the decision to whistleblow, as concerns arise for the security and physical integrity of their respective family members. Law does not provide financial incentives for civil servants to whistleblow. Data reveal that a satisfactory level of trust is missing on organizational indicators such as trust in the responsible officer, protection of data confidentiality and anonymity, across different levels of categories of civil service. In line with the concerns voiced by members of civil service and international standards for whistleblower protection, the following actionable recommendations are proposed to advance the whistleblowing system in Kosovo: 1) Improve the provision of training for members of civil service on whistleblowing legislation, organizational procedures, whistleblower protection, and rights; 2) Establish strategies to support employees for whistleblowing. Such strategies would include programs enabling whistleblowers access to professional services such as stress management, counseling, and legal services; 3) Enhance security measures for the physical integrity of whistleblowers and their respective family members; 4) Establish incentives to encourage whistleblowing, such as financial rewards.

The effect of long-lasting insecticidal nets (LLINs) physical integrity on utilization

Background In Ethiopia, despite improvements in coverage and access, utilization of long-lasting insecticidal nets (LLINs) remains a challenge. Different household-level factors have been identified as associated with LLIN use. However, the contribution of LLIN physical integrity to their utilization is not well investigated and documented. This study aimed to assess the association between the physical integrity of LLINs and their use. Methods This study employed a nested case-control design using secondary data from the Ethiopian LLIN durability monitoring study conducted from May 2015 to June 2018. LLINs not used the night before the survey were identified as cases, while those used the previous night were categorized as controls. The physical integrity of LLINs was classified as no holes, good, acceptable, and torn using the proportionate hole index (pHI). A Generalized Estimating Equation (GEE) model was used to assess and quantify the association between LLIN physical integrity and use. The model specifications included binomial probabilistic distribution, logit link, exchangeable correlation matrix structure, and robust standard errors. The factors included in the model were selected first by fitting binary regression, and then by including all factors that showed statistical significance at P-value less than 0.25 and conceptually relevant variables into the multivariate regression model. Results A total of 5277 observations fulfilled the inclusion criteria. Out of these 1767 observations were cases while the remaining 3510 were controls. LLINs that were in torn physical condition had higher odds (AOR [95% CI] = 1.76 [1.41, 2.19]) of not being used compared to LLINs with no holes. Other factors that showed significant association included the age of the LLIN, sleeping place type, washing status of LLINs, perceptions towards net care and repair, LLIN to people ratio, economic status, and study site. Conclusion and recommendation LLINs that have some level of physical damage have a relatively higher likelihood of not being used. Community members need to be educated about proper care and prevention of LLIN damage to delay the development of holes as long as possible and use available LLINs regularly.

Incidence and consequences of damage to insecticide-treated mosquito nets in Kenya

Background Efforts to improve the impact of long-lasting insecticidal nets (LLINs) should be informed by understanding of the causes of decay in effect. Holes in LLINs have been estimated to account for 7–11% of loss in effect on vectorial capacity for Plasmodium falciparum malaria in an analysis of repeated cross-sectional surveys of LLINs in Kenya. This does not account for the effect of holes as a cause of net attrition or non-use, which cannot be measured using only cross-sectional data. There is a need for estimates of how much these indirect effects of physical damage on use and attrition contribute to decay in effectiveness of LLINs. Methods Use, physical integrity, and survival were assessed in a cohort of 4514 LLINs followed for up to 4 years in Kenya. Flow diagrams were used to illustrate how the status of nets, in terms of categories of use, physical integrity, and attrition, changed between surveys carried out at 6-month intervals. A compartment model defined in terms of ordinary differential equations (ODEs) was used to estimate the transition rates between the categories. Effects of physical damage to LLINs on use and attrition were quantified by simulating counterfactuals in which there was no damage. Results Allowing for the direct effect of holes, the effect on use, and the effect on attrition, 18% of the impact on vectorial capacity was estimated to be lost because of damage. The estimated median lifetime of the LLINs was 2.9 years, but this was extended to 5.7 years in the counterfactual without physical damage. Nets that were in use were more likely to be in a damaged state than unused nets but use made little direct difference to LLIN lifetimes. Damage was reported as the reason for attrition for almost half of attrited nets, but the model estimated that almost all attrited nets had suffered some damage before attrition. Conclusions Full quantification of the effects of damage will require measurement of the supply of new nets and of household stocks of unused nets, and also of their impacts on both net use and retention. The timing of mass distribution campaigns is less important than ensuring sufficient supply. In the Kenyan setting, nets acquired damage rapidly once use began and the damage led to rapid attrition. Increasing the robustness of nets could substantially increase their lifetime and impact but the impact of LLIN programmes on malaria transmission is ultimately limited by levels of use. Longitudinal analyses of net integrity data from different settings are needed to determine the importance of physical damage to nets as a driver of attrition and non-use, and the importance of frequent use as a cause of physical damage in different contexts.

PROGRAMS FOR FORTIFICATION AREA CONSERVATION IN THE NETHERLANDS

The article is aimed at familiarizing readers and specialists with programs for the preservation of fortification territories and methods of using historical landscapes in the Netherlands, the legal aspect of preserving the remains of fortifications and terrain. The project method of integrating the territory into the tourist observation network is considered. Over the past 30 years, the Netherlands has been developing a legal framework in the intersectoral branch for the protection of a significant historical landscape, including existing or lost objects of fortification art of the XVIII and XX centuries. Reasonable protection criteria, a well-formed strategy for systematic use and subordination of the territory prevents violation of the legal framework of the monument and its physical integrity accelerates the process of developing architectural and design proposals for restoring the spatial appearance of the fortification object

Coverage, Usage, Physical Integrity, and Bio-efficacy of Olyset Nets in the Plateau Region, South Benin Following the 2011 Nationwide Distribution

The present study investigated in 8 villages of the Plateau region the coverage, usage, physical integrity, and bio-efficacy of the Olyset nets distributed nationwide by the Benin's National Malaria Control Programme in July 2011. The questionnaire administered as well as the observations made in the households allowed estimating the coverage and usage rates of the 2011 Olyset nets. While their physical integrity was assessed through standard WHO methodology, their bio-efficacy was evaluated through gas chromatography, and WHO cone testing performed with the Kisumu susceptible strain. Mosquito collections through human landing catches (HLCs) were also performed in torn nets to assess if a loss of protection of sleepers occurred as the nets fabric integrity got more damaged. Nine months postdistribution, the coverage and usage rates of the 2011 Olyset nets were 67.4% (95% CI: 65.8–68.9) and 73.3% (95% CI: 70.7–75.8) respectively. About 28% of the 2011 Olyset nets were torn. A drastic drop of the insecticide quantity on the fibers of the nets [from 7.08 µg (95% CI: 5.74–8.42) to 0.2 µg (95% CI: 0.01–0.38)] as well as mortality rates <80% were observed with most nets evaluated. Moreover, the biting rates of An. gambiae s.l. (Diptera: Culicidae) inside torn nets increased in line with their fabric integrity loss. These data support the conclusion that future deployment of nets in the field must be strengthened by community sensitization on their correct use in order to postpone as much as possible appearance of holes and loss of insecticidal activity and encourage repairing of torn nets.