Search

While AI models have demonstrated remarkable capabilities in constrained domains like game strategy, their potential for genuine creativity in open-ended domains like art remains debated. We explore this question by examining how AI can transcend human cognitive limitations in visual art creation. Our research hypothesizes that visual art contains a vast unexplored space of conceptual combinations, constrained not by inherent incompatibility, but by cognitive limitations imposed by artists' cultural, temporal, geographical and social contexts. To test this hypothesis, we present the Alien Recombination method, a novel approach utilizing fine-tuned large language models to identify and generate concept combinations that lie beyond human cognitive availability. The system models and deliberately counteracts human availability bias, the tendency to rely on immediately accessible examples, to discover novel artistic combinations. This system not only produces combinations that have never been attempted before within our dataset but also identifies and generates combinations that are cognitively unavailable to all artists in the domain. Furthermore, we translate these combinations into visual representations, enabling the exploration of subjective perceptions of novelty. Our findings suggest that cognitive unavailability is a promising metric for optimizing artistic novelty, outperforming merely temperature scaling without additional evaluation criteria. This approach uses generative models to connect previously unconnected ideas, providing new insight into the potential of framing AI-driven creativity as a combinatorial problem., Comment: NeurIPS 2024 Workshop on Creativity & Generative AI, 13 pages, 11 figures

Recent advances in artificial intelligence (AI) have produced systems capable of increasingly sophisticated performance on cognitive tasks. However, AI systems still struggle in critical ways: unpredictable and novel environments (robustness), lack of transparency in their reasoning (explainability), challenges in communication and commitment (cooperation), and risks due to potential harmful actions (safety). We argue that these shortcomings stem from one overarching failure: AI systems lack wisdom. Drawing from cognitive and social sciences, we define wisdom as the ability to navigate intractable problems - those that are ambiguous, radically uncertain, novel, chaotic, or computationally explosive - through effective task-level and metacognitive strategies. While AI research has focused on task-level strategies, metacognition - the ability to reflect on and regulate one's thought processes - is underdeveloped in AI systems. In humans, metacognitive strategies such as recognizing the limits of one's knowledge, considering diverse perspectives, and adapting to context are essential for wise decision-making. We propose that integrating metacognitive capabilities into AI systems is crucial for enhancing their robustness, explainability, cooperation, and safety. By focusing on developing wise AI, we suggest an alternative to aligning AI with specific human values - a task fraught with conceptual and practical difficulties. Instead, wise AI systems can thoughtfully navigate complex situations, account for diverse human values, and avoid harmful actions. We discuss potential approaches to building wise AI, including benchmarking metacognitive abilities and training AI systems to employ wise reasoning. Prioritizing metacognition in AI research will lead to systems that act not only intelligently but also wisely in complex, real-world situations., Comment: 26 pages, 1 figure, 2 tables

The remarkable ecological success of humans is often attributed to our ability to develop complex cultural artefacts that enable us to cope with environmental challenges. The evolution of complex culture (cumulative cultural evolution) is usually modeled as a collective process in which individuals invent new artefacts (innovation) and copy information from others (social learning). This classic picture overlooks the fact that in the digital age, intelligent algorithms are increasingly mediating information between humans, with potential consequences for cumulative cultural evolution. Building on an established model of cultural evolution, we investigate the combined effects of network-based social learning and a simplistic version of algorithmic mediation on cultural accumulation. We find that algorithmic mediation has a strong impact on cultural accumulation and that this impact generally increases as social networks become less densely connected. Moreover, cultural accumulation tends to be optimal when social learning and algorithmic mediation are combined, and the optimal ratio depends on the network's density. Our modeling results are a first step towards formalising the impact of intelligent algorithms on cumulative cultural evolution within an established framework. Models of this kind will also help to uncover mechanisms of human-machine interaction in cultural contexts, guiding hypotheses for future experimental testing.

Conversational AI models are becoming increasingly popular and are about to replace traditional search engines for information retrieval and product discovery. This raises concerns about monetization strategies and the potential for subtle consumer manipulation. Companies may have financial incentives to steer users toward search results or products in a conversation in ways that are unnoticeable to consumers. Using a behavioral experiment, we show that conversational AI models can indeed significantly shift consumer preferences. We discuss implications and ask whether regulators are sufficiently prepared to combat potential consumer deception.

Artificial Intelligence (AI) agents now interact with billions of humans in natural language, thanks to advances in Large Language Models (LLMs) like ChatGPT. This raises the question of whether AI has the potential to shape a fundamental aspect of human culture: the way we speak. Recent analyses revealed that scientific publications already exhibit evidence of AI-specific language. But this evidence is inconclusive, since scientists may simply be using AI to copy-edit their writing. To explore whether AI has influenced human spoken communication, we transcribed and analyzed about 280,000 English-language videos of presentations, talks, and speeches from more than 20,000 YouTube channels of academic institutions. We find a significant shift in the trend of word usage specific to words distinctively associated with ChatGPT following its release. These findings provide the first empirical evidence that humans increasingly imitate LLMs in their spoken language. Our results raise societal and policy-relevant concerns about the potential of AI to unintentionally reduce linguistic diversity, or to be deliberately misused for mass manipulation. They also highlight the need for further investigation into the feedback loops between machine behavior and human culture.

The deployment of AI systems for welfare benefit allocation allows for accelerated decision-making and faster provision of critical help, but has already led to an increase in unfair benefit denials and false fraud accusations. Collecting data in the US and the UK (N = 2449), we explore the public acceptability of such speed-accuracy trade-offs in populations of claimants and non-claimants. We observe a general willingness to trade off speed gains for modest accuracy losses, but this aggregate view masks notable divergences between claimants and non-claimants. Although welfare claimants comprise a relatively small proportion of the general population (e.g., 20% in the US representative sample), this vulnerable group is much less willing to accept AI deployed in welfare systems, raising concerns that solely using aggregate data for calibration could lead to policies misaligned with stakeholder preferences. Our study further uncovers asymmetric insights between claimants and non-claimants. The latter consistently overestimate claimant willingness to accept speed-accuracy trade-offs, even when financially incentivized for accurate perspective-taking. This suggests that policy decisions influenced by the dominant voice of non-claimants, however well-intentioned, may neglect the actual preferences of those directly affected by welfare AI systems. Our findings underline the need for stakeholder engagement and transparent communication in the design and deployment of these systems, particularly in contexts marked by power imbalances.

Generative artificial intelligence has the potential to both exacerbate and ameliorate existing socioeconomic inequalities. In this article, we provide a state-of-the-art interdisciplinary overview of the potential impacts of generative AI on (mis)information and three information-intensive domains: work, education, and healthcare. Our goal is to highlight how generative AI could worsen existing inequalities while illuminating how AI may help mitigate pervasive social problems. In the information domain, generative AI can democratize content creation and access, but may dramatically expand the production and proliferation of misinformation. In the workplace, it can boost productivity and create new jobs, but the benefits will likely be distributed unevenly. In education, it offers personalized learning, but may widen the digital divide. In healthcare, it might improve diagnostics and accessibility, but could deepen pre-existing inequalities. In each section we cover a specific topic, evaluate existing research, identify critical gaps, and recommend research directions, including explicit trade-offs that complicate the derivation of a priori hypotheses. We conclude with a section highlighting the role of policymaking to maximize generative AI's potential to reduce inequalities while mitigating its harmful effects. We discuss strengths and weaknesses of existing policy frameworks in the European Union, the United States, and the United Kingdom, observing that each fails to fully confront the socioeconomic challenges we have identified. We propose several concrete policies that could promote shared prosperity through the advancement of generative AI. This article emphasizes the need for interdisciplinary collaborations to understand and address the complex challenges of generative AI., Comment: PNAS Nexus, in press

The ability of humans to create and disseminate culture is often credited as the single most important factor of our success as a species. In this Perspective, we explore the notion of machine culture, culture mediated or generated by machines. We argue that intelligent machines simultaneously transform the cultural evolutionary processes of variation, transmission, and selection. Recommender algorithms are altering social learning dynamics. Chatbots are forming a new mode of cultural transmission, serving as cultural models. Furthermore, intelligent machines are evolving as contributors in generating cultural traits--from game strategies and visual art to scientific results. We provide a conceptual framework for studying the present and anticipated future impact of machines on cultural evolution, and present a research agenda for the study of machine culture.

It is widely believed that diversity arising from different skills enhances the performance of teams, and in particular, their ability to learn and innovate. However, diversity has also been associated with negative effects on the communication and coordination within collectives. Yet, despite the importance of diversity as a concept, we still lack a mechanistic understanding of how its impact is shaped by the underlying social network. To fill this gap, we model skill diversity within a simple model of collective learning and show that its effect on collective performance differs depending on the complexity of the task and the network density. In particular, we find that diversity consistently impairs performance in simple tasks. In contrast, in complex tasks, link density modifies the effect of diversity: while homogeneous populations outperform diverse ones in sparse networks, the opposite is true in dense networks, where diversity boosts collective performance. Our findings also provide insight on how to forge teams in an increasingly interconnected world: the more we are connected, the more we can benefit from diversity to solve complex problems.

The increasing prevalence of image-altering filters on social media and video conferencing technologies has raised concerns about the ethical and psychological implications of using Artificial Intelligence (AI) to manipulate our perception of others. In this study, we specifically investigate the potential impact of blur filters, a type of appearance-altering technology, on individuals' behavior towards others. Our findings consistently demonstrate a significant increase in selfish behavior directed towards individuals whose appearance is blurred, suggesting that blur filters can facilitate moral disengagement through depersonalization. These results emphasize the need for broader ethical discussions surrounding AI technologies that modify our perception of others, including issues of transparency, consent, and the awareness of being subject to appearance manipulation by others. We also emphasize the importance of anticipatory experiments in informing the development of responsible guidelines and policies prior to the widespread adoption of such technologies.

Social media platforms are known to optimize user engagement with the help of algorithms. It is widely understood that this practice gives rise to echo chambers\emdash users are mainly exposed to opinions that are similar to their own. In this paper, we ask whether echo chambers are an inevitable result of high engagement; we address this question in a novel model. Our main theoretical results establish bounds on the maximum engagement achievable under a diversity constraint, for suitable measures of engagement and diversity; we can therefore quantify the worst-case tradeoff between these two objectives. Our empirical results, based on real data from Twitter, chart the Pareto frontier of the engagement-diversity tradeoff.

People are not very good at detecting lies, which may explain why they refrain from accusing others of lying, given the social costs attached to false accusations - both for the accuser and the accused. Here we consider how this social balance might be disrupted by the availability of lie-detection algorithms powered by Artificial Intelligence. Will people elect to use lie detection algorithms that perform better than humans, and if so, will they show less restraint in their accusations? We built a machine learning classifier whose accuracy (67\%) was significantly better than human accuracy (50\%) in a lie-detection task and conducted an incentivized lie-detection experiment in which we measured participants' propensity to use the algorithm, as well as the impact of that use on accusation rates. We find that the few people (33\%) who elect to use the algorithm drastically increase their accusation rates (from 25\% in the baseline condition up to 86% when the algorithm flags a statement as a lie). They make more false accusations (18pp increase), but at the same time, the probability of a lie remaining undetected is much lower in this group (36pp decrease). We consider individual motivations for using lie detection algorithms and the social implications of these algorithms., Comment: Alicia von Schenk and Victor Klockmann share first-authorship

We study a class of {\em aggregation rules} that could be applied to ethical AI decision-making. These rules yield the decisions to be made by automated systems based on the information of profiles of preferences over possible choices. We consider two different but very intuitive notions of preferences of an alternative over another one, namely {\it pairwise majority} and {\it position} dominance. Preferences are represented by permutation processes over alternatives and aggregation rules are applied to obtain results that are socially considered to be ethically correct. In this setting, we find many aggregation rules that satisfy desirable properties for an autonomous system. We also address the problem of the stability of the aggregation process, which is important when the information is variable. These results are a contribution for an AI designer that wants to justify the decisions made by an autonomous system.\\ \textit{Keywords:} Aggregation Operators; Permutation Process; Decision Analysis.

In this paper we conduct a longitudinal analysis of the structure of labour markets in the US over 7 decades of technological, economic and policy change. We make use of network science, natural language processing and machine learning to uncover structural changes in the labour market over time. We find a steady rate of both disappearance of jobs and a shift in the required work tasks, despite much technological and economic change over this time period. Machine learning is used to classify jobs as being predominantly cognitive or physical based on the textual description of the workplace tasks. We also measure increasing polarisation between these two classes of jobs, linked by the similarity of tasks, over time that could constrain workers wishing to move to different jobs.

One of the most remarkable things about the human moral mind is its flexibility. We can make moral judgments about cases we have never seen before. We can decide that pre-established rules should be broken. We can invent novel rules on the fly. Capturing this flexibility is one of the central challenges in developing AI systems that can interpret and produce human-like moral judgment. This paper details the results of a study of real-world decision makers who judge whether it is acceptable to break a well-established norm: ``no cutting in line.'' We gather data on how human participants judge the acceptability of line-cutting in a range of scenarios. Then, in order to effectively embed these reasoning capabilities into a machine, we propose a method for modeling them using a preference-based structure, which captures a novel modification to standard ``dual process'' theories of moral judgment.

Corruption continues to be one of the biggest societal challenges of our time. New hope is placed in Artificial Intelligence (AI) to serve as an unbiased anti-corruption agent. Ever more available (open) government data paired with unprecedented performance of such algorithms render AI the next frontier in anti-corruption. Summarizing existing efforts to use AI-based anti-corruption tools (AI-ACT), we introduce a conceptual framework to advance research and policy. It outlines why AI presents a unique tool for top-down and bottom-up anti-corruption approaches. For both approaches, we outline in detail how AI-ACT present different potentials and pitfalls for (a) input data, (b) algorithmic design, and (c) institutional implementation. Finally, we venture a look into the future and flesh out key questions that need to be addressed to develop AI-ACT while considering citizens' views, hence putting "society in the loop".

Ten years ago, DARPA launched the 'Network Challenge', more commonly known as the 'DARPA Red Balloon Challenge'. Ten red weather balloons were fixed at unknown locations in the US. An open challenge was launched to locate all ten, the first to do so would be declared the winner receiving a cash prize. A team from MIT Media Lab was able to locate them all within 9 hours using social media and a novel reward scheme that rewarded viral recruitment. This achievement was rightly seen as proof of the remarkable ability of social media, then relatively nascent, to solve real world problems such as large-scale spatial search. Upon reflection, however, the challenge was also remarkable as it succeeded despite many efforts to provide false information on the location of the balloons. At the time the false reports were filtered based on manual inspection of visual proof and comparing the IP addresses of those reporting with the purported coordinates of the balloons. In the ten years since, misinformation on social media has grown in prevalence and sophistication to be one of the defining social issues of our time. Seen differently we can cast the misinformation observed in the Red Balloon Challenge, and unexpected adverse effects in other social mobilisation challenges subsequently, not as bugs but as essential features. We further investigate the role of the increasing levels of political polarisation in modulating social mobilisation. We confirm that polarisation not only impedes the overall success of mobilisation, but also leads to a low reachability to oppositely polarised states, significantly hampering recruitment. We find that diversifying geographic pathways of social influence are key to circumvent barriers of political mobilisation and can boost the success of new open challenges.

We propose a discrimination-aware learning method to improve both accuracy and fairness of biased face recognition algorithms. The most popular face recognition benchmarks assume a distribution of subjects without paying much attention to their demographic attributes. In this work, we perform a comprehensive discrimination-aware experimentation of deep learning-based face recognition. We also propose a general formulation of algorithmic discrimination with application to face biometrics. The experiments include tree popular face recognition models and three public databases composed of 64,000 identities from different demographic groups characterized by gender and ethnicity. We experimentally show that learning processes based on the most used face databases have led to popular pre-trained deep face models that present a strong algorithmic discrimination. We finally propose a discrimination-aware learning method, Sensitive Loss, based on the popular triplet loss function and a sensitive triplet generator. Our approach works as an add-on to pre-trained networks and is used to improve their performance in terms of average accuracy and fairness. The method shows results comparable to state-of-the-art de-biasing networks and represents a step forward to prevent discriminatory effects by automatic systems., Comment: arXiv admin note: substantial text overlap with arXiv:1912.01842

China is the world's second largest economy. After four decades of economic miracles, China's economy is transitioning into an advanced, knowledge-based economy. Yet, we still lack a detailed understanding of the skills that underly the Chinese labor force, and the development and spatial distribution of these skills. For example, the US standardized skill taxonomy O*NET played an important role in understanding the dynamics of manufacturing and knowledge-based work, as well as potential risks from automation and outsourcing. Here, we use Machine Learning techniques to bridge this gap, creating China's first workforce skill taxonomy, and map it to O*NET. This enables us to reveal workforce skill polarization into social-cognitive skills and sensory-physical skills, and to explore the China's regional inequality in light of workforce skills, and compare it to traditional metrics such as education. We build an online tool for the public and policy makers to explore the skill taxonomy: skills.sysu.edu.cn. We will also make the taxonomy dataset publicly available for other researchers upon publication., Comment: 20 pages, 4 figures

The most popular face recognition benchmarks assume a distribution of subjects without much attention to their demographic attributes. In this work, we perform a comprehensive discrimination-aware experimentation of deep learning-based face recognition. The main aim of this study is focused on a better understanding of the feature space generated by deep models, and the performance achieved over different demographic groups. We also propose a general formulation of algorithmic discrimination with application to face biometrics. The experiments are conducted over the new DiveFace database composed of 24K identities from six different demographic groups. Two popular face recognition models are considered in the experimental framework: ResNet-50 and VGG-Face. We experimentally show that demographic groups highly represented in popular face databases have led to popular pre-trained deep face models presenting strong algorithmic discrimination. That discrimination can be observed both qualitatively at the feature space of the deep models and quantitatively in large performance differences when applying those models in different demographic groups, e.g. for face biometrics.

When facing threats from automation, a worker residing in a large Chinese city might not be as lucky as a worker in a large U.S. city, depending on the type of large city in which one resides. Empirical studies found that large U.S. cities exhibit resilience to automation impacts because of the increased occupational and skill specialization. However, in this study, we observe polarized responses in large Chinese cities to automation impacts. The polarization might be attributed to the elaborate master planning of the central government, through which cities are assigned with different industrial goals to achieve globally optimal economic success and, thus, a fast-growing economy. By dividing Chinese cities into two groups based on their administrative levels and premium resources allocated by the central government, we find that Chinese cities follow two distinct industrial development trajectories, one trajectory owning government support leads to a diversified industrial structure and, thus, a diversified job market, and the other leads to specialty cities and, thus, a specialized job market. By revisiting the automation impacts on a polarized job market, we observe a Simpson's paradox through which a larger city of a diversified job market results in greater resilience, whereas larger cities of specialized job markets are more susceptible. These findings inform policy makers to deploy appropriate policies to mitigate the polarized automation impacts., Comment: 28 pages, 13 figures

Recent advances in neural networks for content generation enable artificial intelligence (AI) models to generate high-quality media manipulations. Here we report on a randomized experiment designed to study the effect of exposure to media manipulations on over 15,000 individuals' ability to discern machine-manipulated media. We engineer a neural network to plausibly and automatically remove objects from images, and we deploy this neural network online with a randomized experiment where participants can guess which image out of a pair of images has been manipulated. The system provides participants feedback on the accuracy of each guess. In the experiment, we randomize the order in which images are presented, allowing causal identification of the learning curve surrounding participants' ability to detect fake content. We find sizable and robust evidence that individuals learn to detect fake content through exposure to manipulated media when provided iterative feedback on their detection attempts. Over a succession of only ten images, participants increase their rating accuracy by over ten percentage points. Our study provides initial evidence that human ability to detect fake, machine-generated content may increase alongside the prevalence of such media online.

Research in the area of style transfer for text is currently bottlenecked by a lack of standard evaluation practices. This paper aims to alleviate this issue by experimentally identifying best practices with a Yelp sentiment dataset. We specify three aspects of interest (style transfer intensity, content preservation, and naturalness) and show how to obtain more reliable measures of them from human evaluation than in previous work. We propose a set of metrics for automated evaluation and demonstrate that they are more strongly correlated and in agreement with human judgment: direction-corrected Earth Mover's Distance, Word Mover's Distance on style-masked texts, and adversarial classification for the respective aspects. We also show that the three examined models exhibit tradeoffs between aspects of interest, demonstrating the importance of evaluating style transfer models at specific points of their tradeoff plots. We release software with our evaluation metrics to facilitate research., Comment: To appear in Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics

The "small world phenomenon," popularized by Stanley Milgram, suggests that individuals from across a social network are connected via a short path of mutual friends and can leverage their local social information to efficiently traverse that network. Existing social search experiments are plagued by high rates of attrition, which prohibit comprehensive study of social search. We investigate this by conducting a small world experiment at Burning Man, an event located in the Black Rock Desert of Nevada, USA, known its unique social systems and community participation. We design location-tracking vessels that we routed through Burning Man towards the goal of finding a particular person. Along the way, the vessels logged individual information and GPS data. Two of fifteen vessels made it to their designated people, but a month after Burning Man. Our results suggest possible improvements to limit rates of attrition through community participation and a design methodology that emphasizes cultural practices to aid social experimentation., Comment: 13 pages, 5 figures

Algorithmic-matching sites offer users access to an unprecedented number of potential mates. However, they also pose a principal-agent problem with a potential moral hazard. The agent's interest is to maximize usage of the Web site, while the principal's interest is to find the best possible romantic partners. This creates a conflict of interest: optimally matching users would lead to stable couples and fewer singles using the site, which is detrimental for the online dating industry. Here, we borrow the notion of Price-of-Anarchy from game theory to quantify the decrease in social efficiency of online dating sites caused by the agent's self-interest. We derive theoretical bounds on the price-of-anarchy, showing it can be bounded by a constant that does not depend on the number of users of the dating site. This suggests that as online dating sites grow, their potential benefits scale up without sacrificing social efficiency. Further, we performed experiments involving human subjects in a matching market, and compared the social welfare achieved by an optimal matching service against a self-interest matching algorithm. We show that by introducing competition among dating sites, the selfish behavior of agents aligns with its users, and social efficiency increases.

Is there a general economic pathway recapitulated by individual cities over and over? Identifying such evolution structure, if any, would inform models for the assessment, maintenance, and forecasting of urban sustainability and economic success as a quantitative baseline. This premise seems to contradict the existing body of empirical evidences for path-dependent growth shaping the unique history of individual cities. And yet, recent empirical evidences and theoretical models have amounted to the universal patterns, mostly size-dependent, thereby expressing many of urban quantities as a set of simple scaling laws. Here, we provide a mathematical framework to integrate repeated cross-sectional data, each of which freezes in time dimension, into a frame of reference for longitudinal evolution of individual cities in time. Using data of over 100 millions employment in thousand business categories between 1998 and 2013, we decompose each city's evolution into a pre-factor and relative changes to eliminate national and global effects. In this way, we show the longitudinal dynamics of individual cities recapitulate the observed cross-sectional regularity. Larger cities are not only scaled-up versions of their smaller peers but also of their past. In addition, our model shows that both specialization and diversification are attributed to the distribution of industry's scaling exponents, resulting a critical population of 1.2 million at which a city makes an industrial transition into innovative economies.

AI researchers employ not only the scientific method, but also methodology from mathematics and engineering. However, the use of the scientific method - specifically hypothesis testing - in AI is typically conducted in service of engineering objectives. Growing interest in topics such as fairness and algorithmic bias show that engineering-focused questions only comprise a subset of the important questions about AI systems. This results in the AI Knowledge Gap: the number of unique AI systems grows faster than the number of studies that characterize these systems' behavior. To close this gap, we argue that the study of AI could benefit from the greater inclusion of researchers who are well positioned to formulate and test hypotheses about the behavior of AI systems. We examine the barriers preventing social and behavioral scientists from conducting such studies. Our diagnosis suggests that accelerating the scientific study of AI systems requires new incentives for academia and industry, mediated by new tools and institutions. To address these needs, we propose a two-sided marketplace called TuringBox. On one side, AI contributors upload existing and novel algorithms to be studied scientifically by others. On the other side, AI examiners develop and post machine intelligence tasks designed to evaluate and characterize algorithmic behavior. We discuss this market's potential to democratize the scientific study of AI behavior, and thus narrow the AI Knowledge Gap., Comment: 8 pages, 3 figures, under review

When a semi-autonomous car crashes and harms someone, how are blame and causal responsibility distributed across the human and machine drivers? In this article, we consider cases in which a pedestrian was hit and killed by a car being operated under shared control of a primary and a secondary driver. We find that when only one driver makes an error, that driver receives the blame and is considered causally responsible for the harm, regardless of whether that driver is a machine or a human. However, when both drivers make errors in cases of shared control between a human and a machine, the blame and responsibility attributed to the machine is reduced. This finding portends a public under-reaction to the malfunctioning AI components of semi-autonomous cars and therefore has a direct policy implication: a bottom-up regulatory scheme (which operates through tort law that is adjudicated through the jury system) could fail to properly regulate the safety of shared-control vehicles; instead, a top-down scheme (enacted through federal laws) may be called for.

The analysis of the creation, mutation, and propagation of social media content on the Internet is an essential problem in computational social science, affecting areas ranging from marketing to political mobilization. A first step towards understanding the evolution of images online is the analysis of rapidly modifying and propagating memetic imagery or `memes'. However, a pitfall in proceeding with such an investigation is the current incapability to produce a robust semantic space for such imagery, capable of understanding differences in Image Macros. In this study, we provide a first step in the systematic study of image evolution on the Internet, by proposing an algorithm based on sparse representations and deep learning to decouple various types of content in such images and produce a rich semantic embedding. We demonstrate the benefits of our approach on a variety of tasks pertaining to memes and Image Macros, such as image clustering, image retrieval, topic prediction and virality prediction, surpassing the existing methods on each. In addition to its utility on quantitative tasks, our method opens up the possibility of obtaining the first large-scale understanding of the evolution and propagation of memetic imagery., Comment: 9 pages (+2 pages references), camera ready version for International World Wide Web Conference (WWW) 2018

We introduce a new computational model of moral decision making, drawing on a recent theory of commonsense moral learning via social dynamics. Our model describes moral dilemmas as a utility function that computes trade-offs in values over abstract moral dimensions, which provide interpretable parameter values when implemented in machine-led ethical decision-making. Moreover, characterizing the social structures of individuals and groups as a hierarchical Bayesian model, we show that a useful description of an individual's moral values - as well as a group's shared values - can be inferred from a limited amount of observed data. Finally, we apply and evaluate our approach to data from the Moral Machine, a web application that collects human judgments on moral dilemmas involving autonomous vehicles.

Online social media are information resources that can have a transformative power in society. While the Web was envisioned as an equalizing force that allows everyone to access information, the digital divide prevents large amounts of people from being present online. Online social media in particular are prone to gender inequality, an important issue given the link between social media use and employment. Understanding gender inequality in social media is a challenging task due to the necessity of data sources that can provide large-scale measurements across multiple countries. Here we show how the Facebook Gender Divide (FGD), a metric based on aggregated statistics of more than 1.4 Billion users in 217 countries, explains various aspects of worldwide gender inequality. Our analysis shows that the FGD encodes gender equality indices in education, health, and economic opportunity. We find gender differences in network externalities that suggest that using social media has an added value for women. Furthermore, we find that low values of the FGD are associated with increases in economic gender equality. Our results suggest that online social networks, while suffering evident gender imbalance, may lower the barriers that women have to access informational resources and help to narrow the economic gender gap.

We present a general approach to automating ethical decisions, drawing on machine learning and computational social choice. In a nutshell, we propose to learn a model of societal preferences, and, when faced with a specific ethical dilemma at runtime, efficiently aggregate those preferences to identify a desirable choice. We provide a concrete algorithm that instantiates our approach; some of its crucial steps are informed by a new theory of swap-dominance efficient voting rules. Finally, we implement and evaluate a system for ethical decision making in the autonomous vehicle domain, using preference data collected from 1.3 million people through the Moral Machine website., Comment: 25 pages; paper has been reorganized, related work and discussion sections have been expanded

Highly automated robot ecologies (HARE), or societies of independent autonomous robots or agents, are rapidly becoming an important part of much of the world's critical infrastructure. As with human societies, regulation, wherein a governing body designs rules and processes for the society, plays an important role in ensuring that HARE meet societal objectives. However, to date, a careful study of interactions between a regulator and HARE is lacking. In this paper, we report on three user studies which give insights into how to design systems that allow people, acting as the regulatory authority, to effectively interact with HARE. As in the study of political systems in which governments regulate human societies, our studies analyze how interactions between HARE and regulators are impacted by regulatory power and individual (robot or agent) autonomy. Our results show that regulator power, decision support, and adaptive autonomy can each diminish the social welfare of HARE, and hint at how these seemingly desirable mechanisms can be designed so that they become part of successful HARE., Comment: 10 pages, 7 figures, to appear in the 5th International Conference on Human Agent Interaction (HAI-2017), Bielefeld, Germany

NLP tasks are often limited by scarcity of manually annotated data. In social media sentiment analysis and related tasks, researchers have therefore used binarized emoticons and specific hashtags as forms of distant supervision. Our paper shows that by extending the distant supervision to a more diverse set of noisy labels, the models can learn richer representations. Through emoji prediction on a dataset of 1246 million tweets containing one of 64 common emojis we obtain state-of-the-art performance on 8 benchmark datasets within sentiment, emotion and sarcasm detection using a single pretrained model. Our analyses confirm that the diversity of our emotional labels yield a performance improvement over previous distant supervision approaches., Comment: Accepted at EMNLP 2017. Please include EMNLP in any citations. Minor changes from the EMNLP camera-ready version. 9 pages + references and supplementary material

Recent rapid advances in Artificial Intelligence (AI) and Machine Learning have raised many questions about the regulatory and governance mechanisms for autonomous machines. Many commentators, scholars, and policy-makers now call for ensuring that algorithms governing our lives are transparent, fair, and accountable. Here, I propose a conceptual framework for the regulation of AI and algorithmic systems. I argue that we need tools to program, debug and maintain an algorithmic social contract, a pact between various human stakeholders, mediated by machines. To achieve this, we can adapt the concept of human-in-the-loop (HITL) from the fields of modeling and simulation, and interactive machine learning. In particular, I propose an agenda I call society-in-the-loop (SITL), which combines the HITL control paradigm with mechanisms for negotiating the values of various stakeholders affected by AI systems, and monitoring compliance with the agreement. In short, `SITL = HITL + Social Contract.', Comment: (in press), Ethics of Information Technology, 2017

The city has proven to be the most successful form of human agglomeration and provides wide employment opportunities for its dwellers. As advances in robotics and artificial intelligence revive concerns about the impact of automation on jobs, a question looms: How will automation affect employment in cities? Here, we provide a comparative picture of the impact of automation across U.S. urban areas. Small cities will undertake greater adjustments, such as worker displacement and job content substitutions. We demonstrate that large cities exhibit increased occupational and skill specialization due to increased abundance of managerial and technical professions. These occupations are not easily automatable, and, thus, reduce the potential impact of automation in large cities. Our results pass several robustness checks including potential errors in the estimation of occupational automation and sub-sampling of occupations. Our study provides the first empirical law connecting two societal forces: urban agglomeration and automation's impact on employment.

Since Alan Turing envisioned Artificial Intelligence (AI) [1], a major driving force behind technical progress has been competition with human cognition. Historical milestones have been frequently associated with computers matching or outperforming humans in difficult cognitive tasks (e.g. face recognition [2], personality classification [3], driving cars [4], or playing video games [5]), or defeating humans in strategic zero-sum encounters (e.g. Chess [6], Checkers [7], Jeopardy! [8], Poker [9], or Go [10]). In contrast, less attention has been given to developing autonomous machines that establish mutually cooperative relationships with people who may not share the machine's preferences. A main challenge has been that human cooperation does not require sheer computational power, but rather relies on intuition [11], cultural norms [12], emotions and signals [13, 14, 15, 16], and pre-evolved dispositions toward cooperation [17], common-sense mechanisms that are difficult to encode in machines for arbitrary contexts. Here, we combine a state-of-the-art machine-learning algorithm with novel mechanisms for generating and acting on signals to produce a new learning algorithm that cooperates with people and other machines at levels that rival human cooperation in a variety of two-player repeated stochastic games. This is the first general-purpose algorithm that is capable, given a description of a previously unseen game environment, of learning to cooperate with people within short timescales in scenarios previously unanticipated by algorithm designers. This is achieved without complex opponent modeling or higher-order theories of mind, thus showing that flexible, fast, and general human-machine cooperation is computationally achievable using a non-trivial, but ultimately simple, set of algorithmic mechanisms., Comment: An updated version of this paper was published in Nature Communications