.

.

Nadzrul, please introduce yourself shortly.

My name is Nadzrul Hanif, and I am the Co-Founder of Semart, a social impact technology company based in Malaysia. At Semart, we focus on empowering micro and small businesses through digital tools, financial literacy, and business support to help them grow sustainably.

Semart was incubated at SEEd.Lab, where I met my co-founders, Nurliyana Farhana and Dr. H’ng Yin Mun. That experience played a key role in shaping our mission and building the foundation of Semart.

Please describe your latest project in 2-3 sentences.

Our latest project is Semart, a digital ecosystem designed to help micro-entrepreneurs manage and grow their businesses. We provide simple business tools such as POS systems, financial tracking, and data-driven insights, combined with financial literacy and human support, to help underserved businesses improve profitability and long-term resilience.

.

Which social challenge are you most passionate about addressing, and why?

I am most passionate about addressing the economic challenges faced by micro-entrepreneurs, especially those in underserved and rural communities. Many of them work incredibly hard, but lack access to the right tools, knowledge, and opportunities to grow sustainably. I believe empowering these businesses creates a direct ripple effect on families and local communities.

Why did you choose a digital solution instead of a traditional approach?

Digital solutions allow us to scale impact more efficiently and sustainably. Traditional support programs often rely heavily on manual processes and are harder to scale, while technology enables us to provide real-time insights, continuous support, and personalized recommendations to many businesses at once.

How do you involve communities or users in the development of your solution?

We build closely with our users, not just for them. We spend time on the ground listening to micro-business owners, understanding their daily challenges, and continuously improving our solution based on their feedback, behavior, and real-world business needs.

.

What has been your most surprising lesson as a social entrepreneur?

One of the most important lessons I’ve learned is that meaningful innovation starts with deeply understanding real problems. As a social entrepreneur, it is easy to focus on building solutions, but lasting impact comes from listening first, understanding people’s daily challenges, and designing technology that genuinely improves their lives.

I’ve learned that empathy is one of the most important drivers of innovation. The better we understand the people we serve, the more effective and impactful our solutions become.

What emerging digital trend excites you most for social innovation?

I am particularly excited about Artificial Intelligence (AI) because it has the potential to democratize access to knowledge and decision-making. For social innovation, AI can help underserved communities access personalized insights, smarter business recommendations, and better opportunities at a scale that was previously impossible.

.

The post Learning from Social Entrepreneurs: Semart appeared first on ICT4D.at.

.

Please describe your latest project in 2-3 sentences.

Zerobionic builds offline, solar powered humanoid robots that translate STEM lessons into sign language in real time. Made from recycled plastic and co designed with deaf communities, the robots have already reached over 510,000 deaf beneficiaries across the globe.

What makes your solution unique?

We are the first in Africa to build an autonomous humanoid robot that speaks sign language. Unlike imported tech, our robot is built locally, works offline, runs on solar power, and is made from recycled plastic. We are also building the first open source African sign language dataset for STEM, 3.5 million annotated gesture frames, so that any developer, school, or government can build on our work for free.

What advice would you give to other young social entrepreneurs just starting out?

Never give up. The path is long. Keep building. Keep failing. Keep getting back up. The only way to truly lose is to stop. So do not stop. Never give up. Never. Finally, remember: scale is not about more users. It is about more dignity. If your solution does not restore dignity to the people it serves, it is not yet finished.

What is your vision for your project or organisation in the next five years?

In five years, every deaf child across the globe will have access to a sign language robot in their classroom. We will have deployed 1,000 robots across the continent, reaching 3 million deaf learners. Our open source dataset will be the global standard for African sign language AI. And we will have graduated a cohort of deaf engineers, young women and men who learned science through our robots and are now building the next generation of inclusive technology.

.

What emerging digital trend excites you most for social innovation?

Robotics. Its immense potential has only just begun to be tapped. We are moving from software that informs to machines that physically transform the world, classrooms, clinics, and communities. Robotics can reach places no app can, serve people no algorithm can touch. That is where the real revolution lies. And it is already happening.

How do you measure whether your solution is creating real impact?

We track pre and post test STEM comprehension scores across all pilot schools, a 70% improvement. We count every deaf beneficiary reached: over 510,000 and growing. We measure retention: 92% of our youth technicians stay with us. And we collect qualitative feedback from deaf students and teachers, because impact is a deaf girl finally saying, “Now I have a shot at becoming a pilot.” That is the metric that matters.

.

The post Learning from Social entrepreneurs: Zerobionic appeared first on ICT4D.at.

Please describe your latest project in 2-3 sentences.

80% of the worlds disability population live in developing countries. Their education levels are low resulting in high unemployment. Youth4Jobs SwarajAbility is South Asia’s first  exclusive platform for Persons with Disabilities (PwD) which bridges this inclusion gap through accessible, AI-powered skilling, job matching, and livelihood opportunities. By enabling economic participation of youth with disabilities at scale, SwarajAbility unlocks both social impact and economic growth.

.

Which social challenge are you most passionate about addressing, and why?

The social challenge we are most passionate about addressing is the socio-economic exclusion of Persons with Disabilities (PwDs). Despite representing nearly 80 million people in India, over 77% of PwDs remain unemployed or underemployed. ILO states the GDP can increase by 5 to 7% if the disabled are gainfully employed.

Youth4Jobs has studied closely on the field how exclusion from education, skilling, assistive technology, and employment creates a cycle of poverty and dependence. For women with disabilities, these barriers are even greater. There are myths which believe disability is a We believe that livelihood is the most powerful form of inclusion because it creates dignity, independence, and social acceptance.

Our mission is to ensure that disability is no longer a barrier to economic participation.  We have built a complete ecosystem where every person with a disability can learn, earn, and thrive through accessible digital skilling using our LMS, discover assistive technology through a digital marketplace, and find relevant livelihoods through AI-enabled job matching.

Why did you choose a digital solution instead of a traditional approach?

Traditional interventions such as physical training centers, job fairs, and placement drives have created meaningful impact. But they cannot reach the scale required to serve India’s 80 million PwDs. Other challenges faced by many PwDs live in remote locations, face mobility barriers, or lack access to specialized support services.

We therefore chose a digital-first approach to democratize access. SwarajAbility brings together skilling, job opportunities, entrepreneurship, employers, NGOs, academia, and government stakeholders into a single digital ecosystem.

Additionally, SwarajAbility, with large number of youth(supply) and companies (demand) registered, allows us to introduce new products useful to the youth and companies. For example, Our Accessible Learning Management System (LMS), available for anyone who enrols on SwarajAbility, enables PwDs to acquire skills anytime and anywhere through content designed for diverse disabilities. Likewise, Our Assistive Technology and Assistive Devices (AT-AD) digital showcase helps PwDs and companies discover and access products that improve independence and work place productivity.

Technology allows us to move beyond serving thousands to potentially serving millions while ensuring accessibility, personalization, and continuous engagement rather than one-time interventions.

Which technology has been most important for your project, and how do you use it?

Accessibility-focused digital technology combined with Artificial Intelligence has been the most transformative force behind our work.

SwarajAbility leverages AI-powered job matching to connect PwDs with employment opportunities aligned to their skills, preferences, and accessibility needs. The platform also uses AI based location services to generate insights that help employers, governments, and NGOs make better decisions regarding the distribution of PwDs.

Our newer products available to all registered on SwarajAbility like Accessible LMS is built to WCAG standards and incorporates features such as screen-reader compatibility, sign language-enabled content, multilingual support, and simplified navigation. Through the Digital Showcase of AT-AD available on SwarajAbility, youth and companies including caregivers can explore assistive products and solutions that enhance learning, mobility, communication, and workplace productivity.

For us, technology is not just a delivery mechanism – it is an equalizer that removes barriers and creates pathways to economic inclusion for millions of people who have traditionally been excluded from mainstream opportunities.

How do you measure whether your solution is creating real impact?

We believe impact should be measured not by activities completed, but by lives transformed. Therefore, we track outcomes across the entire inclusion journey—from skilling and assistive technology access to employment and income generation.

Some of our key indicators include:

Number of PwDs onboarded – 300,000+

Youth impacted -for livelihoods   1,80,000

Employers onboarded – 3300+

Certificates issued for course completion on e-LMS  : 6600

SwarajAbility has a SROI of 11.1 which means for every 1$ invested, there is a return of 11. Our estimates show $1.6 billion benefits have been unlocked to date through government subsidies, trainings and livelihood linkages. The work helps the country to achieve Sustainable Development Goals.

The work has won several awards like MIT Solve.

What partnerships have been most important for your growth?

Partnerships are the cornerstone of our growth and impact as Disability inclusion requires a collaborative ecosystem.

Employers like JP Morgan, Infosys, Amazon, Accenture, and HDFC Bank have benefited from this diverse talent pool. ILO, CII IBDN (the largest industry consortium); RASCI (Retail Association of India) are partners  Two state governments and two central ministries—Ministry of Social Justice and Empowerment and Ministry of Rural Development—have signed MoUs along with state governments of Karnataka and Maharashtra. Other stakeholders we work with are educational institutions and NGOs.  

The post Learning from Social Entrepreneurs: Youth4Jobs appeared first on ICT4D.at.

.

Daniela, please introduce yourself shortly.

My name is Daniela Alvares Leite, and I am a lawyer, sustainability researcher, and the Founder and CEO of Comida Invisível, a social impact technology company based in São Paulo, Brazil.

I am currently a PhD candidate in Sustainability at the University of São Paulo (USP), researching systemic approaches to reducing food waste.

My work sits at the intersection of technology, public policy, social innovation, and environmental impact, with the goal of building more efficient, inclusive, and sustainable food systems.

Comida Invisível was recognized as a WSA Global Champion in the Environment & Green Energy category.

Please describe your latest project in 2-3 sentences.

Comida Invisível is a digital infrastructure that connects surplus food from restaurants, hotels, supermarkets, schools, hospitals, and food industries to social organizations through geolocation technology, real-time traceability, data analytics, and artificial intelligence.

Beyond technology, we have worked to strengthen the legal and institutional environment for food donation in Brazil. Drawing on my background as a lawyer and social entrepreneur, I actively contributed to discussions that helped shape Brazil’s regulatory framework for food donation, and one of my contributions was incorporated into Brazil’s current National Policy on Food Loss and Food Waste Reduction (Law 15.224/2025).

By transforming food that would otherwise be wasted into measurable social and environmental impact, we have helped generate more than 15 million meals, impacted over 3 million people, and built a network of more than 20,000 mapped social organizations across Brazil.

Our work contributes simultaneously to several Sustainable Development Goals, including Zero Hunger (SDG 2), Sustainable Cities and Communities (SDG 11), Responsible Consumption and Production (SDG 12), Climate Action (SDG 13), and Partnerships for the Goals (SDG 17).

.

What personal experience or observation inspired you to start this project?

The inspiration came from witnessing a contradiction that exists in many societies: enormous quantities of food are wasted every day while millions of people continue to face food insecurity.

I realized that the problem was not food scarcity, but a lack of efficient connections between those with surplus food and those who need it. As a lawyer, I also observed that fear of legal liability was preventing many companies from donating food that was perfectly suitable for consumption.

Technology and legal innovation together offered an opportunity to remove these barriers and create a solution capable of generating impact at scale.

Which technology has been most important for your project, and how do you use it?

Geolocation technology, data analytics, and artificial intelligence are fundamental to our work.

Geolocation allows us to efficiently connect food donors and recipient organizations, while AI and data analytics help optimize donation flows, identify patterns, improve logistics, and generate real-time impact indicators.

Our dashboard provides complete traceability and enables organizations to compare performance across different branches, regions, and operations, helping them reduce waste while increasing social, environmental, and economic value.

How do you measure whether your solution is creating real impact?

We measure impact through social, environmental, and operational indicators.

These include the number of meals generated, kilograms of food recovered, greenhouse gas emissions avoided, organizations supported, and people reached.

To date, our infrastructure has contributed to more than 15 million meals and impacted over 3 million people. Real-time data allows us to continuously monitor outcomes, improve efficiency, and provide transparency to all stakeholders involved.

Which achievement are you most proud of so far?

While I am proud of the awards and international recognition that Comida Invisível has received, including being recognized as a WSA Global Champion, I am most proud of helping create the conditions that allow food donation to happen safely and at scale.

Beyond the technology itself, I am particularly proud that practical lessons learned through Comida Invisível helped inform public policy. Seeing one of my contributions incorporated into Brazil’s current food loss and food waste legislation reinforced my belief that lasting impact requires both innovation and systemic change.

At the same time, we built an infrastructure capable of generating measurable impact. To date, our network has helped generate more than 15 million meals, impacted over 3 million people, and connected thousands of food donors with more than 20,000 mapped social organizations across Brazil.

What makes me especially proud is that we are not addressing a single challenge. By reducing food waste, fighting food insecurity, lowering greenhouse gas emissions, strengthening local communities, and fostering cross-sector collaboration, our work contributes simultaneously to multiple Sustainable Development Goals.

For me, the greatest achievement is proving that technology, public policy, and collaboration can transform inefficiency into impact.

.

What is your strategy for scaling your solution to reach more people?

Our strategy is to position Comida Invisível as infrastructure rather than simply a platform.

We integrate corporations, food service operators, retailers, municipalities, and social organizations into a single ecosystem, creating network effects that expand impact without requiring proportional growth in resources.

By combining technology, partnerships, governance, and standardized processes, we are building a scalable model capable of expanding nationally and internationally.

What advice would you give to young people who want to become social entrepreneurs?

Fall in love with the problem, not with the solution.

When you truly commit yourself to understanding a problem, you begin to see its complexity, its human dimensions, and the systemic factors behind it. This requires curiosity, humility, and a genuine dedication of time, energy, and purpose.

Technology, business models, and solutions may change over time, but a deep commitment to solving a meaningful problem is what sustains long-term impact.

The entrepreneurs who create lasting change are usually those who dedicate not only their minds, but also their hearts, to understanding the challenge they seek to solve.

.

The post Learning from Social Entrepreneurs: Comida Invisível appeared first on ICT4D.at.

Two Days of Big Questions

The 2026 edition was organized around four themes:

1. AI and Health: Digital health systems, access to healthcare, health data governance
2. AI and Work: Future of work, skills, automation in developing economies
3. AI and Social Interactions: Social media, misinformation, digital inclusion, online participation
4. AI and the Environment: Sustainable digitalization, climate justice, environmental impacts of AI

Probably you can already tell, as an AI governance nerd, these are exactly the kinds of topics that are right up my alley. I feel completely in my element when discussing the intersection of technology, ethics, and social impact. Therefore I was excited not only by the presentations themselves, but also by the dicussions with other participants.

AI in healthcare: ethics is becoming institutional, not individual

The conference opened with a strong focus on healthcare. Jennifer Blumenthal-Barby highlighted how AI systems already shape clinical environments through choice architectures and thereby influencing what clinicians notice, how they prioritize information, and which decisions become “easier” to make. Importantly, this is not just a question of user error or overreliance. It is a question of system design. Her message was clear: Governance is an ongoing institutional responsibility and can no longer stop at the level of the physician.

Immediately after the talk, I found myself questioning my life choices: it was 9 a.m., I was facing mathematical formulas, and I was severely under-caffeinated. In his presentation on epistemically-aware AI, Adrià Segarra used mathematics to explain a puzzling phenomenon: why human–AI teams sometimes perform worse than either humans or AI working alone. His core argument: AI is only valuable when it contributes information that humans don’t already have. If an AI system simply tells a clinician what they already know, it adds little. Worse still, people may mistake redundant information for independent confirmation and end up more confident than the evidence warrants. Instead of thinking in terms of “decisions,” we should think in terms of credences — asking what additional information AI actually contributes to our beliefs. An insight that feels increasingly relevant far beyond healthcare.

The next session felt futuristic — though only if you weren’t already familiar with the topic. Janko Munjic discussed criminal culpability in cases where a neuroprosthetic arm causes harm to another person. Where does the responsibility lie: with the user, the clinician, or the manufacturer? In my view, the EU AI Act provides clarity on this matter, yet the conversation remains international in scope.

Mehmet Unver and Iheanyichukwu Henry Ogu presented the hybrid system of UK healthcare AI governance — one built on principles of safety and robustness, with frameworks for transparency and accountability, yet undermined by fragmented oversight structures across institutions.

What emerges is a familiar but unresolved problem: distributed responsibility creates accountability gaps.

Iris Coates McCall tackled the challenge of defining “neurodata” and explored its implications for AI policy and regulation. One statement of hers I will definitely be quoting from now on, because it captures the very issues I keep trying to raise: data is not found, it is produced — we cannot pretend it is neutral.

This connects to a related concern that also blew my mind: whether neurodata should be treated as fundamentally different from other health data. Here, the boundary between human cognition, identity, and digital systems becomes blurred. Neuroethics and AI ethics are increasingly converging. And then came the existential question: how much of neurodata is us? Does it define us as persons? Are we our data?

Nice ideas — but what matters most to me is how to put them into practice.

On the panel on implementing ethical standards in healthcare organizations, Katrina Bramstedt from Roche shared how the company has embedded ethics directly into its AI development process: an “AI Ethics Checklist” guides teams from the very start of every project. What I appreciated most was her pragmatic approach — ethics is treated neither as a one-off compliance exercise nor as an abstract philosophical debate, but as an ongoing conversation, supported by dedicated AI ethicists and rapid ethics consultations.

A lot of the audience liked her analogy that “AI needs a seatbelt.” Regulations are essential, but they can never anticipate every real-world scenario. Organizations therefore need to equip their people with the skills and confidence to make ethical decisions themselves.

From the European Commission, Anca Scortariu outlined efforts to build an “ecosystem of AI excellence and trust”: despite 530 funded projects and a 20× increase in patents, only a quarter of hospitals currently use AI solutions in practice, revealing a significant gap between research and deployment. The missing ingredient is not innovation — it is trust. And building trust takes more than regulation: it requires AI literacy, representative data, transparent governance, and a clear-eyed understanding of both the opportunities and the risks. Her closing message captured the spirit of the discussion perfectly: our goal should not be more AI in healthcare, but better healthcare with AI.

By lunchtime, one thought kept coming back to me: many AI debates are, at their core, debates about accountability. The technology itself is often less complicated than the question of who bears responsibility when something goes wrong.

AI, work, and the quiet transformation of labour

In the afternoon the conference shifted its focus to the future of work and how AI reshapes work itself. Economist Antonin Bergeaud reminded us of the uncertainties, wide discrepancies, and range of possibilities in predictions. Depending on whom you ask, AI could increase GDP by a modest two percent—or trigger almost unimaginable economic growth. The truth is that we simply don’t know yet.

AI could end scarcity, end humanity or boost trend growth by 0.2%

One reason is surprisingly simple: we don’t even have a common understanding of what counts as AI adoption. Employees are already using ChatGPT and other AI tools every day, often without their employers even knowing. AI is spreading quietly, one personal subscription at a time.

That resonated with the theme of the next session: “Reconfiguring Creative Labour.” While the headlines suggest otherwise, most people aren’t waking up to find themselves replaced by AI. They’re finding themselves working differently. Writing differently. Researching differently. Solving problems differently. In reality, AI doesn’t lead to replacement, but selective adaptation. That shift requires a different skill set and the capacity to reframe problems effectively.

Tom van Nuenen explored how generative AI influences trust between colleagues. What happens when we can no longer tell how much effort someone invested, whether a text genuinely reflects their voice, or even whether they stand behind what they produced? Different professions value different signals of authenticity: scientists care deeply about provenance and citations, but designers often care about creativity and effort. Across all professions, however, one thing remains constant: trust. AI doesn’t just change what we produce – it changes how we evaluate each other’s work.

Quote that shows how workers mask AI usage

The next session from (hilarious) Narmadha Kamalakannanv shifted the focus from technology to people. Organizations are rapidly investing in AI, yet very few are asking whether their employees actually feel ready for this transformation. One sentence resonated strongly with me: “Psychological safety is not a soft skill—it is the foundation for successful AI transformation.” We spend enormous resources measuring productivity, efficiency, and return on investment, but rarely measure confidence, fear, or readiness. Perhaps successful AI adoption depends just as much on listening to employees as it does on deploying new tools.

The panel discussion at the end of the day brought many of these ideas together. Mar Carpanelli from LinkedIn noted that working on one of the world’s biggest datasets on job updates tells “a boring story” – there is currently no evidence of widespread job displacement. Instead, LinkedIn’s data points to augmentation: new roles such as data annotators are emerging, and new skills like prompt engineering are gaining traction. The skills in highest demand are abilities that signal competence in working alongside AI: problem formulation, coordination, and influence. When it comes to workforce training, the real differentiator may not be any specific curriculum but rather a growth mindset about agility, flexibility, and the capacity for ongoing problem-solving. Everyone is asking how many jobs will be affected, how quickly, and at what cost. Far fewer are asking who benefits and who bears the burden. Men are roughly twice as likely as women to engage with AI development tools, raising serious questions about equity in who shapes this transition.

Yann Ferguson from LaborIA introduced the concept of “AI debt.” We often celebrate the time AI saves us, but rarely ask what we might be losing in return. Learning debt, because we let AI think for us. Knowledge debt, because we ask ChatGPT instead of our colleagues. Identity debt, because we slowly lose ownership over our work. Even sovereignty debt, because dependence on a handful of AI providers gradually limits our autonomy.

AI and social interactions: What happens when machines become companions?

The second day opened with a topic that felt surprisingly personal. After spending so much time discussing governance frameworks, organizational responsibility, and the future of work, the conversation shifted to something much more fundamental: how AI is changing the way we relate to each other—and to ourselves.

One of the opening talks by Matthias Scheutz explored what he called the social lure of generative AI. His central argument was both simple and unsettling: what appears social is not necessarily social. Large language models are becoming increasingly good at simulating empathy, encouragement, and companionship. They remember details, adapt to our communication style, and often tell us exactly what we want to hear. Yet these interactions are fundamentally one-sided. Unlike human relationships, there is no mutual understanding, vulnerability, or genuine reciprocity. Drawing on Sherry Turkle‘s work, Scheutz highlighted the inherent dangers of unidirectional emotional bonds between humans and social robots, with children being especially vulnerable. He pointed to Replika as a stark example: users who engage with it most frequently are also the loneliest, raising the specter of what he termed “AI psychosis.” Real-world harms have followed—deaths linked to chatbots are already documented on Wikipedia—making the case for safeguards and regulation urgent rather than abstract.

Sherry Turkle’s “artificial intimacy”

Chiara Marcoccia shifted the lens from individual psychology to societal dynamics, asking whether we are entering a period of human-AI coevolution. Her framework emphasized that human-AI interaction produces network effects that ripple outward, altering not just individual behavior but the fabric of society.

One presentation that particularly caught my attention challenged a topic that often appears in AI discussions: value alignment. We frequently hear that AI should reflect human values, but the researchers demonstrated how difficult that actually is in practice. Arno Libert tested whether leading AI models can make consistent moral decisions. His team constructed agentic dilemmas: scenarios involving a chemical spill, a smart-home upsell targeting an elderly person, and fintech transparency. They presented moral dilemmas to different language models, slightly rephrasing the scenarios each time. The results were striking. Even small changes in wording produced entirely different moral judgments. None of the tested models consistently expressed coherent ethical reasoning across all scenarios. It was a powerful reminder that today’s AI systems are exceptionally good at generating plausible language, but that should not be mistaken for stable moral competence.

A panel moderated among Nataliya Kosmyna, Maria Melchior, and Sylvie Delacroix brought these threads together in a wide-ranging discussion on interacting with LLMs. Kosmyna drew a provocative parallel, likening TikTok’s influence on youth to the fentanyl crisis – a comparison she did not make lightly, given the documented mental health deterioration among young people. Melchior added nuance: the harms are not specifically AI-driven but rather compound existing vulnerabilities: young people already in difficult circumstances, with addicted parents or precarious home environments are disproportionately affected. Crucially, she emphasized that AI’s influence extends to physical health, an area that receives far less attention than psychological harm. Delacroix reframed the conversation around how these systems communicate uncertainty. Confidence scores and contextual uncertainty are not merely technical features, they also carry ethical weight. As cooperation with AI tools deepens, she warned, we tend to reduce our critical thinking, and this transformation is already reshaping the way we learn.

AI and the environment: the material footprint of intelligence

The environmental sessions grounded the discussion in physical infrastructure: energy use, data centers, and resource extraction.

Critical headlines against AI

ICT4D research will recognize this: digital solutions often redistribute environmental costs unevenly.

Ontological Communitarism and Environmental Justice in AI Governance

Impressions

Taken together, the conference pointed to three broader shifts:

1. AI ethics is becoming institutional governance, not individual morality
2. AI systems are reshaping epistemic structures, not just outputs
3. The core challenge is no longer adoption, but distribution — of power, costs, and benefits

Many of the same questions long asked in development contexts are now appearing in high-income settings:

• Who benefits from digital transformation?
• Who bears the risks?
• Who is included in governance?
• Who gets to define “trustworthy AI”?

A recurring challenge in our work in ICT4D is that new technologies are often presented as solutions to social and economic development. Yet despite unprecedented technological progress oever the last decades, inequalities continue to deepen within societies. And with AI the same issues increase exponentially while it is being implemented before the social impacts are properly understood. It is crucial to lead these discussions and to lead them now. Personally I missed questions (and critique) of power, accountability, and governance. As an academic conference there is a lack of translation of the knowledge into the practical real-life scenarios. This is the nature of an academic conference – but a few business leaders were present that shared their insights and best practices.

Digital technologies have never been an end in themselves. They should ultimately improve people’s lives, strengthen communities, and expand opportunities. AI should be no different. The goal isn’t simply to make work faster or more automated—it is to make work more meaningful, more inclusive, and more human.

The conference repeatedly returned to a question that resonates strongly with ICT4D: not whether AI can create value, but for whom, under what conditions, and at whose expense.

Group photo at PCAIDE 2026

The post The Paris Conference on AI & Digital Ethics appeared first on ICT4D.at.

.

Please introduce yourself.
We are Salome Margvelashvili and Ion Dormenco, working on WSA 2025 winning digital solution Vote Monitor within Commit Global and Code for Romania. Salome is a Product Manager based in Bucharest, and Ion is a Senior Technology Officer based in Iași. Together we work on civic tech tools that support election observation and transparency in different countries.

Please describe your latest project in 2-3 sentences.
Vote Monitor, is a mobile and web platform that helps election observers report what happens at polling stations in real time to NGOs and therefore fully digitalizes election observation practices. This transforms election observation from a delayed documentation to real life democratic protection as speed and accuracy of the information collected by NGOs gives them a possibility to see the patterns and systemic violations as the elections are still ongoing and react to them.

.

Which social challenge are you most passionate about addressing, and why?

We care a lot about the gap between what happens during elections and when it actually becomes visible to the public and to oversight organizations.
In many places, election observation is still a very manual process: observers write their observations down on paper, sometimes take photos separately, and only after polls close do NGOs start compiling everything into reports. That means serious issues that happen in the morning or midday often only become public knowledge hours later, when it’s no longer possible to react.
The problem is not just speed for the sake of speed. it’s that timing changes impact. If a violation happens, the ability to respond, document it properly, or escalate it legally is often time-sensitive. So delays don’t just slow reporting, they can reduce accountability.
What we are trying to change is that fundamental delay. We want election observation to be something that happens in real time where signals from the field are visible quickly enough that NGOs can still act while election day is ongoing, not only after it is already over.

Why did you choose a digital solution instead of a traditional approach?

Because the traditional approach is simply too slow for how elections work today. Paper-based reporting, or even fragmented tools like messaging apps, creates delays and loses structure. Important signals get buried or arrive too late. Vote Monitor allows us to standardize reporting,
speed up communication, and make sure critical issues get flagged by the system, especially during the most sensitive hours of election day.

How do you involve communities or users in the development of your solution?

Vote Monitor is constantly evolving together with the NGOs that use it. Every organization has its own methodology, so we don’t build for one fixed workflow. Instead, we design it to be flexible and adaptable across different contexts. We work closely with partners during deployments, collect feedback after every election, and continuously improve the tool so it stays useful in very different environmentsm from small missions to large international coalitions. Overall, Vote Monitor has been used as a data collection tool in 39 election rounds across three continents, by 8500 of observers and over 100 NGOs – and always provided for free.

How do you balance social impact and financial sustainability?

We keep Vote Monitor free for all NGOs, because we don’t want financial barriers to limit access to election observation tools that aid democratic processes. Our sustainability comes from public funding, grants, and institutional support from partners who believe in democratic transparency. This model allows us to focus on impact and trust, not monetization, while still being able to maintain and scale the platform.

.

What partnerships have been most important for your growth?

Partnerships are critical in election observation because the field is very sensitive and can easily be politicized. We are very careful about who we work with and always aim to support legitimate, independent civil society organizations.
One of our key partners in Europe is EPDE (European Platform for Democratic Elections) and their network of member organizations. They have been especially important for us, not only in helping us reach credible election observation groups across different countries, but also in strengthening due diligence around who we support. EPDE’s research on fake election observation missions has been particularly valuable in this context, as it helps the whole ecosystem better identify and avoid organizations that are not genuinely independent.
We also trust the networks like RedOIE in Latin America and ELOG in Africa, who play a similar role in connecting us with established organizations on the ground. For us, these partnerships are about trust, accountability, and making sure the tool is used in a responsible and democratic way.

The post Learning from Social Entrepreneurs: appeared first on ICT4D.at.

.

Dustin, please introduce yourself.
My name is Dustin Klepper. I am co-founder and managing director of neuraflow, a GovTech company based in Bremerhaven on the German North Sea coast. Together with my co-founder Pascal Nobereit I build AI solutions that make public administration more digital, more accessible and more citizen-friendly.

Please describe your latest project in 2-3 sentences.
Our flagship is neurapolis, an AI platform for municipalities. It addresses a very practical problem: citizens struggle to find the right information and reach the right office, while administrations are overloaded and increasingly short on staff. neurapolis combines a website assistant, a telephone voicebot, on-site terminals and internal tools for employees, all drawing on the same curated official sources, so people get reliable answers around the clock and staff are freed up for the cases that genuinely need a human. Our larger goal is simple: to make the work of public administration more transparent and accessible, so that government opens doors for people instead of building hurdles.

.

What personal experience or observation inspired you to start this project?

We kept seeing the same picture in German towns. Residents waiting on hold or standing in line for something as simple as registering a car or booking an appointment, while the people behind the counter were buried under repetitive questions. At the same time, public administrations are about to lose a large share of their workforce to retirement. It felt almost unfair that the technology to ease this already existed everywhere in the private sector but had not yet reached the public services that everyone depends on. Closing that gap is what got us started.

Which social challenge are you most passionate about addressing, and why?

Equal and barrier-free access to public services, and just as importantly, making the work of the state transparent and understandable. A chatbot or a voicebot is only truly inclusive if everyone can use it, including people with disabilities, older citizens and people who are not fluent in German, which is why, to our knowledge, we are currently the only provider on the market that fully meets the German BITV 2.0 accessibility standard. But inclusion is about more than the interface. When residents can actually understand what their administration decides and easily reach the service they need, trust in public institutions grows. Public administration is the one service no one can opt out of, so it has to be open to everyone.

Which technology has been most important for your project, and how do you use it?

Clearly large language models, but the way we use them matters more than the model itself. Everything runs on sovereign infrastructure hosted in Germany, without any hyperscaler, and on request even fully on premise, because public data has to stay under public control. The real engineering sits in anchoring the AI strictly in the official information each municipality provides, so the assistant does not invent answers but reliably reflects what the administration actually communicates, across every channel from the website to the phone line. Internally, the same approach makes council records and administrative knowledge far easier to search, which makes the work of the state more transparent for the people inside it too.

Who benefits most from your solution, and what have you learned from working with them?

Two groups benefit most. Citizens get correct answers immediately, at any hour and in plain language, and municipal staff are relieved of the flood of routine requests so they can concentrate on the cases that really need a person. The biggest lesson from working with more than 200 municipalities is that trust is everything in the public sector. People forgive a private chatbot that occasionally guesses, but they will not forgive a city assistant that gives wrong information. Reliability and transparency beat clever features every single time.

.

What change do you hope your solution will create in five years?

In five years I want talking to your local administration to feel as natural and immediate as any good digital service you already use today, with no opening hours, no waiting loops and no dead ends, and fully accessible to everyone. Our vision is an administration that opens doors instead of building hurdles, where state action is transparent and within reach for every citizen. If a resident in a small village receives the same quality of service as someone in a large city, simply because both administrations run on a sovereign and citizen-friendly AI layer, then we have achieved what we set out to do.

The post Learning from Social Entrepreneurs: neuropolis appeared first on ICT4D.at.

Fabián, please introduce yourself.
My name is Fabián Curzio, and I am the Co-Founder and CEO of ELDES, an AI-powered platform that makes sign language learning accessible through real-time feedback and artificial intelligence. Originally from Uruguay and currently based in Spain, I lead the company’s vision, product development, and strategic growth. Together with our team, we are working to break communication barriers and create a more inclusive world for deaf and hard-of-hearing communities.

What`s ELDES exactly?
ELDES is an AI-powered platform that makes sign language learning accessible, scalable, and engaging. We address the global shortage of sign language teachers and the barriers many people face when trying to learn, using computer vision and artificial intelligence to provide real-time feedback through a standard device camera. This allows anyone to practice and improve their signing skills anytime, anywhere, at their own pace.

What personal experience or observation inspired you to start this project?

ELDES was born from a very personal experience. My brother Martín and I both have low vision, so we grew up understanding firsthand the barriers that people with disabilities often face in education, employment, and everyday life.

As we learned more about the deaf community, we discovered a significant gap: millions of people wanted to learn sign language, but access to qualified teachers was limited, expensive, and difficult to scale. We realized that artificial intelligence could help bridge that gap by making sign language education accessible to anyone, anywhere. That conviction inspired us to create ELDES and dedicate our work to building a more inclusive society through technology.

Who benefits most from your solution, and what have you learned from working with them?

Our solution benefits a wide range of users, including deaf and hard-of-hearing communities, educators, public servants, healthcare professionals, customer service teams, and anyone interested in communicating more inclusively.

One of the most important lessons we have learned is that accessibility is not only about technology – it is about confidence. Many people want to learn sign language but lack the time, access, or opportunity to attend traditional courses. By making learning flexible, interactive, and available anywhere, we help more people take the first step toward meaningful inclusion and communication.

Can you share a story of a user whose life was changed through your project?

One story that deeply impacted us came from a teacher in Spain who completed our sign language course through a public education program. After finishing the training, she told us that she finally felt confident communicating directly with deaf students in her classroom, without always relying on an interpreter or third party.

What stayed with us was not only the new skill she had acquired, but the change in her mindset. She explained that learning sign language helped her better understand the daily barriers faced by deaf people and made her a stronger advocate for inclusion. Stories like hers remind us that every new person who learns sign language helps create a more accessible society.

How do you involve communities or users in the development of your solution?

The deaf community has been involved in ELDES from the very beginning. Our content is created and validated together with deaf sign language users, interpreters, educators, and accessibility experts to ensure both linguistic accuracy and cultural relevance.

We also maintain a continuous feedback loop with our users and institutional partners. Their experiences, suggestions, and challenges directly influence our product roadmap, helping us improve the learning experience and develop new features that respond to real-world needs.

Which achievement are you most proud of so far?

One of the achievements we are most proud of is demonstrating that technology can dramatically expand access to sign language education at scale. In just a few years, ELDES has enabled thousands of people to start learning sign language through an accessible, self-paced experience powered by AI.

We are also honored to have received international recognition, including winning the award for the World’s Most Innovative EdTech at the Global EdTech Awards, being named a WSA Global Champion, and being selected as a Tech for Change startup at VivaTech 2026. Beyond the awards, what matters most to us is seeing institutions, companies, and public organizations adopt our solution to build more inclusive environments for deaf communities.

What is your strategy for scaling your solution to reach more people?

Our strategy is to combine technology, partnerships, and localization. Because ELDES is a digital platform powered by AI, we can scale efficiently and reach large numbers of users without the limitations of traditional classroom-based training.

We work closely with governments, educational institutions, and companies to expand access to sign language learning, while adapting our content to the specific sign languages and cultural contexts of different countries. This approach allows us to increase our impact globally while ensuring that the learning experience remains relevant and accessible to local communities.

What advice would you give to young people who want to become social entrepreneurs?

Start by deeply understanding a real problem and the people affected by it. The most impactful social ventures are built alongside communities, not just for them. Spend time listening, validating assumptions, and making sure the problem is worth solving before focusing on the solution.

At the same time, remember that impact and sustainability must go hand in hand. A social enterprise can only create long-term change if it has a viable business model. Be patient, stay resilient, and don’t be afraid to start small – many of the most meaningful innovations begin with a simple idea and a genuine desire to make a difference.

The post Learning from Social Entrepreneurs: ELDES appeared first on ICT4D.at.

Philips, please introduce yourself.
My name is Philips Olajide-Philips. I am the Founder & CEO of KorinAI, an AI-powered music creation platform. I am based in Lagos, Nigeria.

What’s your latest project with social impact?
My latest project is KorinAI, an AI music platform that enables artists, creators, and businesses to generate studio-quality songs, jingles, and soundtracks in minutes without needing a physical studio. It solves the high cost and inaccessibility of music production by using AI models trained on African sounds and languages. Users can simply input text or ideas and instantly produce ready-to-use music content for personal, commercial, or media use.

What personal experience or observation inspired you to start this project?

My journey started at the intersection of music and technology. Having studied music technology and worked with creators across Africa, I saw firsthand how many talented artists could not afford studio production. At the same time, I noticed that emerging AI tools did not understand African languages, voices, or musical styles. That inspired me to build KorinAI – to make music creation accessible while preserving African culture in the age of AI.

Why did you choose a digital solution instead of a traditional approach?

Traditional music production is expensive, time-consuming, and often inaccessible to millions of creators. A digital AI-powered platform allows us to reach users anywhere with an internet connection, dramatically reduce costs, and empower people to create professional-quality music in minutes rather than weeks.

Who benefits most from your solution, and what have you learned from working with them?

Emerging artists, content creators, churches, small businesses, and media producers benefit the most from KorinAI. Working with these communities taught us that creativity is everywhere, but access is not. Many people have powerful ideas and stories to share; they simply need affordable tools to bring them to life.

Which achievement are you most proud of so far?

I am most proud that KorinAI is helping preserve African cultural identity through technology. Beyond growing to over 1,100 users and generating more than 1,200 songs, our work has been recognized internationally, including recognition through the World Summit Awards and acceptance into the NVIDIA Inception Program.

What advice would you give to young people who want to become social entrepreneurs?

Start with a real problem, not just an idea. Spend time understanding the people you want to serve and build solutions with them, not for them. Most importantly, be persistent. Creating impact takes time, but if you stay focused on solving meaningful problems, opportunities and support will follow.

The post Learning from Social Entrepreneurs: KorinAI appeared first on ICT4D.at.

Wednesday: Conference Opening and Setting the Stage 

Going into the conference, we were open-minded and looking forward to the conversations, building on the momentum from the ICT4D Conference in Ghana in 2024. The opening keynotes largely focused on the future of digital development, exploring how emerging technologies, data, and digital public infrastructure are shaping development practice, while also raising important questions around inclusion, governance, sustainability, and local ownership.

As the conference progressed, one theme became increasingly difficult to ignore: Artificial Intelligence (AI). We did not expect an avalanche of discussions around AI, yet it featured prominently throughout the sessions. What stood out to us, however, was that the conversations were far more nuanced than simply promoting the latest tools and technologies. Instead, discussions centred on responsible adoption, governance, inclusion, capacity building, and understanding when AI is or is not the appropriate solution. This was particularly refreshing, as many AI conversations often focus on capabilities rather than consequences. The panel discussions and technical demonstrations on the use of AI in humanitarian contexts were particularly enriching. Examples ranged from predicting disaster-prone areas to supporting refugee response efforts and improving the efficiency of humanitarian operations. The discussions highlighted both the opportunities and the realities of deploying AI in environments where resources, infrastructure, and human capacity often remain constrained.

As researchers interested in digital inclusion, we found ourselves reflecting on a recurring question: how do we ensure that the communities most likely to benefit from these technologies are not the very ones left behind by their adoption? Rather than presenting AI as the answer to every challenge, many of the discussions acknowledged the importance of context, readiness, and human-centred implementation.

Thursday: Shifting Digital Power, voice technologies and Tech Demonstrations

Thursday began with opening keynote sessions that set the stage for conversations around digital transformation, inclusion, and responsible technology adoption.

Following the tea break, Jack and I facilitated breakouts in a workshop on shifting AI digital power hosted by Matt Haikin. The session explored issues around South-to-South cooperation (Jack Odunga), Control (Anne Muchiri), Capacity (Nyagaki Gichia), and Practice (Mala Ram). Different facilitators and participants approached the sub-topics from various perspectives, creating a rich discussion around who benefits from AI, who controls it, and who is often left out of the conversation.

One of the first activities involved understanding participants’ perceptions and experiences with AI. We explored whether participants understood what AI is and whether they had begun integrating it into their workflows. What emerged was a fascinating spectrum. While some participants were highly familiar with AI and actively experimenting with it, others had little awareness of how it could be applied in their work and no clear pathways to begin adoption. What surprised us was not the level of enthusiasm for AI, but the unevenness of adoption. This highlighted that conversations about AI adoption cannot assume a common starting point, particularly within development contexts. 

This observation also connected closely with discussions around South-to-South cooperation (led by Jack) and control (led by Anne) within AI ecosystems. Participants challenged the assumption that innovation must always flow from the Global North, emphasizing the growing role of knowledge exchange and locally developed solutions across Africa, Asia, and Latin America. Equally important were questions of control : Who develops the technologies? Who owns the data? Who sets the rules? And ultimately, who benefits from the value created? These discussions reinforced the importance of ensuring that communities and countries are not merely consumers of AI technologies, but active participants in shaping their development and use. 

After our session, we attended a session on “Voice as the Next Frontier for Inclusive Conversational AI in Low-Resource Language Contexts”. Questions emerged around the complexity of voice compared to text-based interactions. While voice can lower barriers for many users, particularly those with literacy challenges, it also introduces new layers of complexity around language, context, cultural nuances, and interpretation. The day concluded with technical demonstrations showcasing a range of digital tools, including conversational applications and platforms designed to support communication and information tracking.

One message that resonated strongly throughout the day was that AI is not a silver bullet. Before deciding on any technology, we must first understand the problem. Only then can we determine whether AI is the appropriate solution, or whether a simpler, more traditional approach may be more effective.

Friday: Data, Practice, and Scale

Friday was packed with engaging presentations and discussions.

The day opened with keynote presentations that continued the conversation around digital transformation, responsible innovation, and the realities of implementing technology in development and humanitarian contexts. Subsequent sessions explored data collection tools and methodologies, including practical demonstrations of platforms such as CommCare and Kobo Collect. While the technologies themselves were important, the conversations increasingly centered on data quality, governance, and responsibility rather than simply data collection. The conference later concluded with additional technical demonstrations that showcased how organizations are leveraging technology to improve programme delivery and decision-making.

Key Reflections

One of the strongest themes throughout the conference was the inclusion of communities and young people in technology conversations. Rather than designing solutions in isolation, there was a growing emphasis on ensuring that affected communities participate in shaping the systems intended to serve them.

What also became clear is that conversations around digital transformation have matured significantly. The focus is no longer solely on deploying technology but on understanding whether organizations and ecosystems are actually ready to adopt and sustain these technologies. Questions around data governance, security, infrastructure, skills, and organizational readiness featured prominently throughout the discussions.

Several recurring themes emerged:

• Data governance remains central. Collecting the right data, ensuring its quality, and using it responsibly are increasingly recognized as foundational requirements.
• Interoperability continues to be a major challenge, particularly when organizations and governments struggle to share information effectively.
• Political will remains a critical factor in enabling meaningful data sharing and collaboration.
• Partnerships are essential, especially in the humanitarian sector, where there is a growing desire to move beyond endless pilots and towards sustainable scaling.
• Questions around data centers, hosting, energy consumption, and water usage are becoming increasingly important as digital infrastructure expands.
• Bias, inclusion, and fairness remain critical considerations in AI development and deployment.
• Organizations are beginning to think more seriously about staff capacity, reskilling, and redesigning workflows to accommodate emerging technologies.
• Discussions around data sovereignty, AI sovereignty, local language technologies, and small language models reflected a growing interest in ensuring that countries and communities retain agency over their digital futures.

A Shift Towards Responsible Digital Transformation

Perhaps the biggest takeaway was the growing sense of maturity in digital transformation conversations. A few years ago, many discussions around technology seemed driven primarily by the promise of innovation and efficiency. Today, there appears to be a stronger recognition that human-centred values, inclusion, ethics, governance, and sustainability must be integral to technology development. As researchers working on digital inclusion and equitable access to technology, this shift strongly resonated with our own work and experiences. 

The conversations repeatedly reinforced the idea that AI is simply one tool among many. It may be the right solution in some situations, but not all. What matters most is understanding the problem, the context, and the people involved before deciding on the intervention.

Overall, we left encouraged by the increasing emphasis on responsibility in digital transformation. Rather than pursuing technology for its own sake or focusing solely on profit and efficiency, there seems to be a growing commitment to ensuring that technological progress remains grounded in human needs, local realities, and long-term societal value.

The post Reflections from the ICT4D 2026 Conference: Beyond the AI Hype appeared first on ICT4D.at.