Areas of Inquiry
Our research spans the intersection of AI technology and human need. These focus areas inform the products we build and the problems we choose to solve.
Research-Driven Development
At SoftenAI, research isn't separate from product development—it's foundational to it. We believe that building products for the masses requires deep understanding of the contexts, constraints, and opportunities that shape how people interact with technology.
Our research directly informs product decisions, from understanding labor market dynamics for our micro-task platforms to studying how people learn AI concepts for our education initiatives.
Future of Work
The nature of work is fundamentally changing. We research how AI transforms labor markets, creates new forms of employment, and opens pathways to economic opportunity—particularly in regions traditionally underserved by technology.
Micro-Task Economies
Understanding how bite-sized work creates income opportunities for millions who lack access to traditional employment.
Skills for AI Era
Identifying which skills remain valuable as AI capabilities expand, and how workers can develop them.
Platform Design
Building work platforms that maximize human potential rather than extract value from vulnerable workers.
Natural Language Understanding
Language is the primary interface between humans and AI. We advance NLP capabilities for multilingual contexts, focusing on languages and dialects often overlooked by mainstream research—serving billions, not just the English-speaking world.
Low-Resource Languages
Developing models that work for languages with limited training data, from Bengali to Swahili.
Conversational AI
Building dialogue systems that understand cultural context, not just grammar and vocabulary.
Content Understanding
Scaling content moderation and analysis across languages while respecting cultural nuances.
Human-AI Interaction
Great AI is invisible—it enhances human capability without creating friction. We study how people actually interact with AI systems to design experiences that feel intuitive, accessible, and trustworthy across all demographics.
Trust Calibration
Helping users understand when to rely on AI recommendations and when to apply their own judgment.
Universal Accessibility
Designing for first-time smartphone users and experts alike, across literacy and ability levels.
Error Recovery
Creating graceful experiences when AI gets things wrong, maintaining user confidence.
AI for Emerging Markets
Building AI for emerging markets requires understanding constraints that don't exist in developed economies—intermittent connectivity, shared devices, varying literacy levels, and diverse payment ecosystems. We build AI that works within these realities.
Offline-First Design
AI systems that function without constant connectivity and sync gracefully when networks return.
Shared Device Patterns
Understanding how families and communities share devices, and designing for multi-user contexts.
Local Payment Integration
Connecting AI products with mobile money, local banking, and informal payment systems.
AI Education
AI literacy shouldn't be a privilege. We explore how to make AI concepts accessible to everyone—from students in under-resourced schools to professionals looking to upskill—meeting learners where they are.
Community Learning
Peer-driven education models that scale without requiring expert instructors everywhere.
Practical Skills
Moving beyond theory to hands-on AI capabilities people can apply immediately.
Visual Learning
Interactive approaches that make abstract AI concepts tangible and understandable.
Responsible AI
Responsible AI isn't a checkbox—it's a continuous practice woven into how we build. We develop practical frameworks for identifying bias, assessing societal impact, and ensuring our systems reflect the values we claim.
Bias Detection
Practical methods for finding and mitigating bias in production systems, not just research papers.
Impact Assessment
Measuring how our AI systems affect communities, especially vulnerable populations.
Algorithmic Accountability
Creating transparency about how decisions are made and who is responsible for outcomes.
Career AI
Career development is often opaque and unequal—who you know matters more than what you know. We research how AI can democratize access to career guidance, surface hidden opportunities, and create fair pathways to advancement.
Skills Matching
Connecting what people can do with opportunities they might not know exist.
Learning Pathways
Personalized recommendations for skill development based on career goals.
Fair Hiring
Reducing bias in how candidates are discovered, evaluated, and selected.
Recommendation Systems
Recommendations shape what people see, learn, and buy. We build systems that are not only accurate but fair—promoting discovery and diversity rather than trapping users in filter bubbles of their own past behavior.
Diverse Discovery
Recommendations that expand horizons rather than narrowing them over time.
Explainability
Helping users understand why something was recommended and how to adjust.
Cold Start Solutions
Providing useful recommendations for new users with no history to learn from.
Cognitive AI
The best AI augments human cognition rather than replacing it. We research how to model human reasoning patterns and build AI that complements how people actually think—respecting cognitive limits while extending capabilities.
Decision Support
AI that helps humans make better decisions without taking agency away.
Cognitive Load
Reducing mental burden through intelligent defaults and progressive disclosure.
Mental Models
Aligning AI behavior with how users expect systems to work.
AI Privacy
Privacy isn't optional—especially for vulnerable populations who have the most to lose. We develop techniques for privacy-preserving machine learning that enable powerful AI without compromising the trust users place in us.
Federated Learning
Training models across devices without centralizing sensitive personal data.
Differential Privacy
Mathematical guarantees that individual data cannot be extracted from AI systems.
Synthetic Data
Generating realistic training data that protects real user information.
See Research in Action
Our research directly shapes the products we build. Explore how these insights translate into real-world solutions.
Social Behavior
AI systems operate within social contexts. We study human social dynamics, community formation, and online behavior patterns to build AI that strengthens connections rather than fragmenting communities.
Trust Formation
How trust develops in digital spaces and how AI can facilitate rather than undermine it.
Community Health
Detecting and addressing toxic dynamics before they destroy online communities.
Positive Reinforcement
Designing systems that encourage helpful behavior and meaningful connection.