Using Wikipedia for conflict forecasting


Article by Christian Oswald and Daniel Ohrenhofer: “How can we improve our ability to predict conflicts? Scholars have struggled with this question for a long time. However, as a discipline, and especially over the last two decades, political science has made substantial progress. In general, what we need to improve predictions are advances in data and methodology. Data advances involve both improving the quality of existing data and developing new data sources. We propose a new data source for conflict forecasting efforts: Wikipedia.

The number of country page views indicates international salience of, or interest in, a country. Meanwhile, the number of changes to a country page indicate political controversy between opposing political views.

We took part in the Violence Early-Warning System’s friendly competition to predict changes in battle-related deaths. In our work, we evaluate our findings with out-of-sample predictions using held-out, previously unseen data, and true forecasts into the future. We find support for the predictive power of country page views, whereas we do not for page changes…

Globally available data, updated monthly, are ideal for (near) real-time forecasting. However, many commonly used data sources are available only annually. They are updated once a year, often with considerable delay.

Some of these variables, such as democracy or GDP, tend to be relatively static over time. Furthermore, many data sources face the problem of missing values. These occur when it is not possible to find reliable data for a variable for a given country.

Wikipedia is updated in real time, unlike many commonly used data sources, which may update only annually and with considerable delay

More recent data sources such as Twitter, images or text as data, or mobile phone data, often do not provide global coverage. What’s more, collecting and manipulating data from such sources is typically computationally and/or financially costly. Wikipedia provides an alternative data source that, to some extent, overcomes many of these limitations…(More)”.

Unsustainable Alarmism


Essay by Taylor Dotson: “Covid is far from the only global challenge we see depicted as a cataclysm in the making. In 1968, Paul Ehrlich predicted impending famine and social collapse driven by overpopulation. He compared the threat to a ticking bomb — the “population bomb.” And the claim that only a few years remain to prevent climate doom has become a familiar refrain. The recent film Don’t Look Up, about a comet barreling toward Earth, is obviously meant as an allegory for climate catastrophe.

But catastrophism fails to capture the complexities of problems that play out over a long time scale, like Covid and climate change. In a tornado or a flood, which are not only undeniably serious but also require immediate action to prevent destruction, people drop political disputes to do what is necessary to save lives. They bring their loved ones to higher ground. They stack sandbags. They gather in tornado shelters. They evacuate. Covid began as a flood in early 2020, but once a danger becomes long and grinding, catastrophism loses its purchase, and more measured public thinking is required.

Even if the extension of catastrophic rhetoric to longer-term and more complex problems is well-intentioned, it unavoidably implies that something is morally or mentally wrong with the people who fail to take heed. It makes those who are not already horrified, who do not treat the crisis as an undeniable, act-now-or-never calamity, harder to comprehend: What idiot wouldn’t do everything possible to avert catastrophe? This kind of thinking is why global challenges are no longer multifaceted dilemmas to negotiate together; they have become conflicts between those who recognize the self-evident truth and those who have taken flight from reality….(More)”.

Non-human humanitarianism: when ‘AI for good’ can be harmful


Paper by Mirca Madianou: “Artificial intelligence (AI) applications have been introduced in humanitarian operations in order to help with the significant challenges the sector is facing. This article focuses on chatbots which have been proposed as an efficient method to improve communication with, and accountability to affected communities. Chatbots, together with other humanitarian AI applications such as biometrics, satellite imaging, predictive modelling and data visualisations, are often understood as part of the wider phenomenon of ‘AI for social good’. The article develops a decolonial critique of humanitarianism and critical algorithm studies which focuses on the power asymmetries underpinning both humanitarianism and AI. The article asks whether chatbots, as exemplars of ‘AI for good’, reproduce inequalities in the global context. Drawing on a mixed methods study that includes interviews with seven groups of stakeholders, the analysis observes that humanitarian chatbots do not fulfil claims such as ‘intelligence’. Yet AI applications still have powerful consequences. Apart from the risks associated with misinformation and data safeguarding, chatbots reduce communication to its barest instrumental forms which creates disconnects between affected communities and aid agencies. This disconnect is compounded by the extraction of value from data and experimentation with untested technologies. By reflecting the values of their designers and by asserting Eurocentric values in their programmed interactions, chatbots reproduce the coloniality of power. The article concludes that ‘AI for good’ is an ‘enchantment of technology’ that reworks the colonial legacies of humanitarianism whilst also occluding the power dynamics at play…(More)”.

The digitalisation of social protection before and since the onset of Covid-19: opportunities, challenges and lessons


Paper by the Overseas Development Institute: “…discusses the main opportunities and challenges associated with digital social protection, drawing on trends pre-Covid and since the onset of the pandemic. It offers eight lessons to help social protection actors capitalise on technology’s potential in a risk-sensitive manner.

  • The response to Covid-19 accelerated the trend of increasing digitalisation of social protection delivery.
  • Studies from before and during the pandemic suggest that well-used technology holds potential to enhance provision for some service users, and played a notable role in rapid social protection expansion during Covid-19. It may also help reduce leakage or inclusion errors, lower costs and support improvements in programme design.
  • However, unless designed and implemented with careful mitigating measures, digitalisation may in some cases do more harm than good. Key concerns relate to potential risks and challenges of exclusion, protection and privacy violations, ‘technosolutionism’ and obscured transparency and accountability.
  • Ultimately, technology is a tool, and its outcomes depend on the needs it is expected to meet, the goals it is deployed to pursue, and the specific ways in which it is designed and implemented…(More)”.

Data scientists are using the most annoying feature on your phones to save lives in Ukraine


Article by Bernhard Warner: “In late March, five weeks into Russia’s war on Ukraine, an international team of researchers, aid agency specialists, public health experts, and data nerds gathered on a Zoom call to discuss one of the tragic by-products of the war: the refugee crisis.

The numbers discussedweregrim. The United Nations had just declared Ukraine was facing the biggest humanitarian crisis to hit Europe since World War II as more than 4 million Ukrainians—roughly 10% of the population—had been forced to flee their homes to evade Russian President Vladimir Putin’s deadly and indiscriminate bombing campaign. That total has since swelled to 5.5 million, the UN estimates.

What the aid specialists on the call wanted to figure out was how many Ukrainian refugees still remained in the country (a population known as “internally displaced people”) and how many had crossed borders to seek asylum in the neighboring European Union countries of Poland, Slovakia, and Hungary, or south into Moldova. 

Key to an effective humanitarian response of this magnitude is getting accurate and timely data on the flow of displaced people traveling from a Point A danger zone to a Point B safe space. And nobody on the call, which was organized by CrisisReady, an A-team of policy experts and humanitarian emergency responders, had anything close to precise numbers.

But they did have a kind of secret weapon: mobility data.

“The importance of mobility data is often overstated,” Rohini Sampoornam Swaminathan, a crisis specialist at Unicef, told her colleagues on the call. Such anonymized data—pulled from social media feeds, geolocation apps like Google Maps, cell phone towers and the like—may not give the precise picture of what’s happening on the ground in a moment of extreme crisis, “but it’s valuable” as it can fill in points on a map. ”It’s important,” she added, “to get a picture for where people are moving, especially in the first days.”

Ukraine, a nation of relatively tech-savvy social media devotees and mobile phone users, is rich in mobility data, and that’s profoundly shaped the way the world sees and interprets the deadly conflict. The CrisisReady group believes the data has an even higher calling—that it can save lives.

Since the first days of Putin’s bombing campaign, various international teams have been tapping publicly available mobility data to map the refugee crisis and coordinate an effective response. They believe the data can reveal where war-torn Ukrainians are now, and even where they’re heading. In the right hands, the data can provide local authorities the intel they need to get essential aid—medical care, food, and shelter—to the right place at the right time…(More)”

Data sharing between humanitarian organisations and donors


Report by Larissa Fast: “This report investigates issues related to data sharing between humanitarian actors and donors, with a focus on two key questions:

  • What formal or informal frameworks govern the collection and sharing of disaggregated humanitarian data between humanitarian actors and donors?
  • How are these frameworks and the related requirements understood or perceived by humanitarian actors and donors?

Drawing on interviews with donors and humanitarians about data sharing practices and examination of formal documents, the research finds that, overall and perhaps most importantly, references to ‘data’ in the context of humanitarian operations are usually generic and lack a consistent definition or even a shared terminology. Complex regulatory frameworks, variability among donor expectations, both among and within donor governments (e.g., at the country or field/headquarters levels), and among humanitarian experiences of data sharing all complicate the nature and handling of data sharing requests. Both the lack of data literacy and the differing perceptions of operational data management risks exacerbate many issues related to data sharing and create inconsistent practice (see full summary of findings in Table 3).

More specifically, while much formal documentation about data sharing between humanitarians and donors is available in the public domain, few contain explicit policies or clauses on data sharing, instead referring only to financial or compliance data and programme reporting requirements. Additionally, the justifications for sharing disaggregated humanitarian data are framed most often in terms of accountability, compliance, efficiency, and programme design. Most requests for data are linked to monitoring and compliance, as well as requests for data as ‘assurances’. Even so, donors indicated that although they request detailed/disaggregated data, they may not have the time, or human and/or technical capacity to deal with it properly. In general, donor interviewees insisted that no record level data is shared within their governments, but only aggregated or in low or no sensitivity formats….(More)”.

Solferino 21: Warfare, Civilians and Humanitarians in the Twenty-First Century


Book by Hugo Slim: “War is at a tipping point: we’re passing from the age of industrial warfare to a new era of computerised warfare, and a renewed risk of great-power conflict. Humanitarian response is also evolving fast—‘big aid’ demands more and more money, while aid workers try to digitalise, preparing to meet ever-broader needs in the long, big wars and climate crisis of the future. 

This book draws on the founding moment of the modern Red Cross movement—the 1859 Battle of Solferino, a moment of great change in the nature of conflict—to track the big shifts already underway, and still to come, in the wars and war aid of our century. Hugo Slim first surveys the current landscape: the tech, politics, law and strategy of warfare, and the long-term transformations ahead as conflict goes digital. He then explains how civilians both suffer and survive in today’s wars, and how their world is changing. Finally, he critiques today’s humanitarian system, citing the challenges of the 2020s.   

Inspired by Henri Dunant’s seminal humanitarian text, Solferino 21 alerts policymakers to the coming shakeup of the military and aid professions, illuminating key priorities for the new century. Humanitarians, he warns, must adapt or fail….(More)”.

Selected Readings on Digital Self-Determination for Migrants


By Uma Kalkar, Marine Ragnet, and Stefaan Verhulst

Digital self-determination (DSD) is a multidisciplinary concept that extends self-determination to the digital sphere. Self-determination places humans (and their ability to make ‘moral’ decisions) at the center of decision-making actions. While self-determination is considered as a jus cogens rule (i.e. a global norm), the concept of digital self-determination came only to light in the early 2010s as a result of the increasing digitization of most aspects of society. 

While digitalization has opened up new opportunities for self-expression and communication for individuals across the globe, its reach and benefits have not been evenly distributed. For instance, migrants and refugees are particularly vulnerable to the deepening inequalities and power structures brought on by increased digitization, and the subsequent datafication. Further, non-traditional data, such as social media and telecom data, have brought great potential to improve our understanding of the migration experience and patterns of mobility that can provide more targeted migration policies and services yet it also has brought new concerns related to the lack of agency to determine how the data is being used and who determines the migration narrative.

These selected readings look at DSD in light of the growing ubiquity of technology applications and specifically focus on their impacts on migrants. They were produced to inform the first studio on DSD and migration co-hosted by the Big Data for Migration Alliance and the International Digital Self Determination Network. The readings are listed in alphabetical order.

These readings serve as a primer to offer base perspectives on DSD and its manifestations, as well as provide a better understanding of how migration data is managed today to advance or hinder life for those on the move. Please alert us of any other publication we should include moving forward.

Berens, Jos, Nataniel Raymond, Gideon Shimshon, Stefaan Verhulst, and Lucy Bernholz. “The Humanitarian Data Ecosystem: the Case for Collective Responsibility.” Stanford Center for Philanthropy and Civil Society, 2017.

  • The authors explore the challenges to, and potential solutions for, the responsible use of digital data in the context of international humanitarian action. Data governance is related to DSD because it oversees how the information extracted from an individual—understood by DSD as an extension of oneself in the digital sphere—is handled.
  • They argue that in the digital age, the basic service provision activities of NGOs and aid organizations have become data collection processes. However, the ecosystem of actors is “uncoordinated” creating inefficiencies and vulnerabilities in the humanitarian space.
  • The paper presents a new framework for responsible data use in the humanitarian domain. The authors advocate for data users to follow three steps: 
  1. “[L]ook beyond the role they take up in the ‘data-lifecycle’ and consider previous and following steps and roles;
  2. Develop sound data responsibility strategies not only to prevent harm to their own operations but also to other organizations in the ‘data-lifecycle;’ and, 
  3. Collaborate with and learn from other organizations, both in the humanitarian field and beyond, to establish broadly supported guidelines and standards for humanitarian data use.”

Currion, Paul. “The Refugee Identity.Caribou Digital (via Medium), March 13, 2018.

  • Developed as part of a DFID-funded initiative, this essay outlines the Data Requirements for Service Delivery within Refugee Camps project that investigated current data standards and design of refugee identity systems.
  • Currion finds that since “the digitisation of aid has already begun…aid agencies must therefore pay more attention to the way in which identity systems affect the lives and livelihoods of the forcibly displaced, both positively and negatively.” He argues that an interoperable digital identity for refugees is essential to access financial, social, and material resources while on the move but also to tap into IoT services.
  • However, many refugees are wary of digital tracking and data collection services that could further marginalize them as they search for safety. At present, there are no sector-level data standards around refugee identity data collection, combination, and centralization. How can regulators balance data protection with government and NGO requirements to serve refugees in the ways they want to uphold their DSD?
  • Currion argues that a Responsible Data approach, as opposed to a process defined by a Data Minimization principle, provides “useful guidelines” but notes that data responsibility “still needs to be translated into organizational policy, then into institutional processes, and finally into operational practice. He further adds that “the digitization of aid, if approached from a position that empowers the individual as much as the institution, offers a chance to give refugees back their voices.”

Decker, Rianne, Paul Koot, S. Ilker Birbil, Mark van Embden Andres. “Co-designing algorithms for governance: Ensuring responsible and accountable algorithmic management of refugee camp supplies” Big Data and Society, April 2022. 

  • While recent literature has looked at the negative impacts of big data and algorithms in public governance, claiming they may reinforce existing biases and defy scrutiny by public officials, this paper argues that designing algorithms with relevant government and society stakeholders might be a way to make them more accountable and transparent. 
  • It presents a case study of the development of an algorithmic tool to estimate the populations of refugee camps to manage the delivery of emergency supplies. The algorithms included in this tool were co-designed with relevant stakeholders. 
  • This may provide a way to uphold DSD by  contributing to the “accountability of the algorithm by making the estimations transparent and explicable to its users.”
  • The authors found that the co-design process enabled better accuracy and responsibility and fostered collaboration between partners, creating a suitable purpose for the tool and making the algorithm understandable to its users. This enabled algorithmic accountability. 
  • The authors note, however, that the beneficiaries of the tools were not included in the design process, limiting the legitimacy of the initiative. 

European Migration Network. “The Use of Digitalisation and Artificial Intelligence in Migration Management.” EMN-OECD Inform Series, February 2022.

  • This paper explores the role of new digital technologies in the management of migration and asylum, focusing specifically on where digital technologies, such as online portals, blockchain, and AI-powered speech and facial recognition systems are being used across Europe to navigate the processes of obtaining visas, claiming asylum, gaining citizenship,  and deploying border control management. 
  • Further, it points to friction between GDPR and new technologies like blockchain—which by decision does not allow for the right to be forgotten—and potential workarounds, such as two-step pseudonymisation.
  • As well, it highlights steps taken to oversee and open up data protection processes for immigration. Austria, Belgium, and France have begun to conduct Data Protection Impact Assessments; France has a portal that allows one to request the right to be forgotten; Ireland informs online service users on how data can be shared or used with third-party agencies; and Spain outlines which personal data are used in immigration as per the Registry Public Treatment Activities.
  • Lastly, the paper points out next steps for policy development that upholds DSD, including universal access and digital literacy, trust in digital systems, willingness for government digital transformations, and bias and risk reduction.

Martin, Aaron, Gargi Sharma, Siddharth Peter de Souza, Linnet Taylor, Boudewijn van Eerd, Sean Martin McDonald, Massimo Marelli, Margie Cheesman, Stephan Scheel, and Huub Dijstelbloem. “Digitisation and Sovereignty in Humanitarian Space: Technologies, Territories and Tensions.” Geopolitics (2022): 1-36.

  • This paper explores how digitisation and datafication are reshaping sovereign authority, power, and control in humanitarian spaces.
  • Building on the notion that technology is political, Martin et al. discuss three cases where digital tools powered by partnerships between international organizations and NGOs and private firms such as Palantir and Facebook have raised concerns for data to be “repurposed” to undermine national sovereignty and distort humanitarian aims with for-profit motivations.
  • The authors draw attention to how cyber dependencies threaten international humanitarian organizations’ purported digital sovereignty. They touch on the tensions between national and digital sovereignty and self-governance.
  • The paper further argues that the rise of digital technologies in the governance of international mobility and migration policies “has all kinds of humanitarian and security consequences,” including (but not limited to) surveillance, privacy infringement, profiling, selection, inclusion/exclusion, and access barriers. Specifically, Scheel introduces the notion of function creep—the use of digital data beyond initially defined purposes—and emphasizes its common use in the context of migration as part “of the modus operandi of sovereign power.”

McAuliffe, Marie, Jenna Blower, and Ana Beduschi. “Digitalization and Artificial Intelligence in Migration and Mobility: Transnational Implications of the COVID-19 Pandemic.” Societies 11, no. 135 (2021): 1-13.

  • This paper critically examines the implications of intensifying digitalization and AI for migration and mobility systems in a post- COVID transnational context. 
  • The authors first situate digitalization and AI in migration by analyzing its uptake throughout the Migration Cycle, i.e. to verify identities and visas, “enable “smart” border processing,” and understand travelers’ adherence to legal frameworks. It then evaluates the current challenges and opportunities to migrants and migration systems brought about by deepening digitalization due to COVID-19. For example, contact tracing, infection screening, and quarantining procedures generate increased data about an individual and are meant, by design, to track and trace people, which raises concerns about migrants’ safety, privacy, and autonomy.
  • This essay argues that recent changes show the need for further computational advances that incorporate human rights throughout the design and development stages, “to mitigate potential risks to migrants’ human rights.” AI is severely flawed when it comes to decision-making around minority groups because of biased training data and could further marginalize vulnerable populations and intrusive data collection for public health could erode the power of one’s universal right to privacy. Leaving migrants at the mercy of black-box AI systems fails to uphold their right to DSD because it forces them to relinquish their agency and power to an opaque system.

Ponzanesi, Sandra. “Migration and Mobility in a Digital Age: (Re)Mapping Connectivity and Belonging.” Television & New Media 20, no. 6 (2019): 547-557.

  • This article explores the role of new media technologies in rethinking the dynamics of migration and globalization by focusing on the role of migrant users as “connected” and active participants, as well as “screened” and subject to biometric datafication, visualization, and surveillance.
  • Elaborating on concepts such as “migration” and “mobility,” the article analyzes the paradoxes of intermittent connectivity and troubled belonging, which are seen as relational definitions that are always fluid, negotiable, and porous.
  • It states that a city’s digital infrastructures are “complex sociotechnical systems” that have a functional side related to access and connectivity and a performative side where people engage with technology. Digital access and action represent areas of individual and collective manifestations of DSD. For migrants, gaining digital access and skills and “enacting citizenship” are important for resettlement. Ponzanesi advocates for further research conducted both from the bottom-up that leans on migrant experiences with technology to resettle and remain in contact with their homeland and a top-down approach that looks at datafication, surveillance, digital/e-governance as a part of the larger technology application ecosystem to understand contemporary processes and problems of migration.

Remolina, Nydia, and Mark James Findlay. “The Paths to Digital Self-Determination — A Foundational Theoretical Framework.” SMU Centre for AI & Data Governance Research Paper No. 03 (2021): 1-34.

  • Remolina and Findlay stress that self-determination is the vehicle by which people “decide their own destiny in the international order.” Decision-making ability powers humans to be in control of their own lives and excited to pursue a set of actions. Collective action, or the ability to make decisions as a part of a group—be it based on ethnicity, nationality, shared viewpoints, etc.—further motivates oneself.
  • The authors discuss how the European Union and European Court of Human Rights’ “principle of subsidiarity” aligns with self-determination because it advocates for power to be placed at the lowest level possible to preserve bottom-up agency with a “reasonable level of efficiency.” In practice, the results of subsidiarity have been disappointing.
  • The paper provides examples of indigenous populations’ fight for self-determination, offline and online. Here, digital self-determination refers to the challenges indigenous peoples face in accessing growing government uses of technology for unlocking innovative solutions because of a lack of physical infrastructure due to structural and social inequities between settler and indigenous communities.
  • Understanding self-determination—and by extension, digital self-determination as a human right, the report investigates how autonomy, sovereignty, the legal definition of a ‘right,’ inclusion, agency, data governance, data ownership, data control, and data quality.
  • Lastly, the paper presents a foundational theoretical framework that goes beyond just protecting personal data and privacy. Understanding that DSD “cannot be detached from duties for responsible data use,” the authors present a collective and individual dimension to DSD. They extend the individual dimension of DSD to include both my data and data about me that can be used to influence a person’s actions through micro-targeting and nudge techniques. They update the collective dimension of DSD to include the views and influences of organizations, businesses, and communities online and call for a better way of visualizing the ‘social self’ and its control over data.

Ziebart, Astrid, and Jessica Bither. “AI, Digital Identities, Biometrics, Blockchain: A Primer on the Use of Technology in Migration Management.” Migration Strategy Group on International Cooperation and Development, June 2020.

  • Ziebart and Bither note the implications of increasingly sophisticated use of technology and data collection by governments with respect to their citizens. They note that migrants and refugees “often are exposed to particular vulnerabilities” during these processes and underscore the need to bring migrants into data gathering and use policy conversations.  
  • The authors discuss the promise of technology—i.e., to predict migration through AI-powered analyses, employ technologies to reduce friction in the asylum-seeking processes, and the power of digital identities for those on the move. However, they stress the need to combine these tools with informational self-determination that allows migrants to own and control what data they share and how and where the data are used.
  • The migration and refugee policy space faces issues of “tech evangelism,” where technologies are being employed just because they exist, rather than because they serve an actual policy need or provide an answer to a particular policy question. This supply-driven policy implementation signals the need for more migrant voices to inform policymakers on what tools are actually useful for the migratory experience. In order to advance the digital agency of migrants, the paper offers recommendations for some of the ethical challenges these technologies might pose and ultimately advocates for greater participation of migrants and refugees in devising technology-driven policy instruments for migration issues.

On-the-go interesting resources 

  • Empowering Digital Self-Determination, mediaX at Stanford University: This short video presents definitions of DSD, and digital personhood, identity, and privacy and an overview of their applications across ethics, law, and the private sector.
  • Digital Self-Determination — A Living Syllabus: This syllabus and assorted materials have been created and curated from the 2021 Research Sprint run by the Digital Asia Hub and Berkman Klein Center for Internet Society at Harvard University. It introduces learners to the fundamentals of DSD across a variety of industries to enrich understanding of its existing and potential applications.
  • Digital Self-Determination Wikipedia Page: This Wikipedia page was developed by the students who took part in the Berkman Klein Center research sprint on digital self-determination. It provides a comprehensive overview of DSD definitions and its key elements, which include human-centered design, robust privacy mandates and data governance, and control over data use to give data subjects the ability to choose how algorithms manipulate their data for autonomous decision-making.
  • Roger Dubach on Digital Self-Determination: This short video presents DSD in the public sector and the dangers of creating a ‘data-protected’ world, but rather on understanding how governments can efficiently use data and protect privacy. Note: this video is part of the Living Syllabus course materials (Digital Self-Determination/Module 1: Beginning Inquiries).

Co-designing algorithms for governance: Ensuring responsible and accountable algorithmic management of refugee camp supplies


Paper by Rianne Dekker et al: “There is increasing criticism on the use of big data and algorithms in public governance. Studies revealed that algorithms may reinforce existing biases and defy scrutiny by public officials using them and citizens subject to algorithmic decisions and services. In response, scholars have called for more algorithmic transparency and regulation. These are useful, but ex post solutions in which the development of algorithms remains a rather autonomous process. This paper argues that co-design of algorithms with relevant stakeholders from government and society is another means to achieve responsible and accountable algorithms that is largely overlooked in the literature. We present a case study of the development of an algorithmic tool to estimate the populations of refugee camps to manage the delivery of emergency supplies. This case study demonstrates how in different stages of development of the tool—data selection and pre-processing, training of the algorithm and post-processing and adoption—inclusion of knowledge from the field led to changes to the algorithm. Co-design supported responsibility of the algorithm in the selection of big data sources and in preventing reinforcement of biases. It contributed to accountability of the algorithm by making the estimations transparent and explicable to its users. They were able to use the tool for fitting purposes and used their discretion in the interpretation of the results. It is yet unclear whether this eventually led to better servicing of refugee camps…(More)”.

Google is using AI to better detect searches from people in crisis


Article by James Vincent: “In a personal crisis, many people turn to an impersonal source of support: Google. Every day, the company fields searches on topics like suicide, sexual assault, and domestic abuse. But Google wants to do more to direct people to the information they need, and says new AI techniques that better parse the complexities of language are helping.

Specifically, Google is integrating its latest machine learning model, MUM, into its search engine to “more accurately detect a wider range of personal crisis searches.” The company unveiled MUM at its IO conference last year, and has since used it to augment search with features that try to answer questions connected to the original search.

In this case, MUM will be able to spot search queries related to difficult personal situations that earlier search tools could not, says Anne Merritt, a Google product manager for health and information quality.

“MUM is able to help us understand longer or more complex queries like ‘why did he attack me when i said i dont love him,’” Merrit told The Verge. “It may be obvious to humans that this query is about domestic violence, but long, natural-language queries like these are difficult for our systems to understand without advanced AI.”

Other examples of queries that MUM can react to include “most common ways suicide is completed” (a search Merrit says earlier systems “may have previously understood as information seeking”) and “Sydney suicide hot spots” (where, again, earlier responses would have likely returned travel information — ignoring the mention of “suicide” in favor of the more popular query for “hot spots”). When Google detects such crisis searches, it responds with an information box telling users “Help is available,” usually accompanied by a phone number or website for a mental health charity like Samaritans.

In addition to using MUM to respond to personal crises, Google says it’s also using an older AI language model, BERT, to better identify searches looking for explicit content like pornography. By leveraging BERT, Google says it’s “reduced unexpected shocking results by 30%” year-on-year. However, the company was unable to share absolute figures for how many “shocking results” its users come across on average, so while this is a comparative improvement, it gives no indication of how big or small the problem actually is.

Google is keen to tell you that AI is helping the company improve its search products — especially at a time when there’s a building narrative that “Google search is dying.” But integrating this technology comes with its downsides, too.

Many AI experts warn that Google’s increasing use of machine learning language models could surface new problems for the company, like introducing biases and misinformation into search results. AI systems are also opaque, offering engineers restricted insight into how they come to certain conclusions…(More)”.