Belem-LLM: Where the Amazon Meets the Algorithm
Teaching AI Ethics Before It Speaks
Belem-LLM is a Fulbright-born initiative led by Dr. Mark Lennon connecting Para, Brazil to Pittsburgh, Pennsylvania and beyond. We are building a bi-directional, multilingual, ethically trained language model that listens, learns, and reflects our best human values.
Check out some of our AI education Apps
AI Magic Trick
AI Ethics Are Being Taught Too Late
Most discussions focus on how AI is used — not how it's built. Generative models mirror biases baked into their data, and without ethically balanced corpora, we perpetuate inequity, not insight.
  • Educators fear "GPT = Great Plagiarism Tool"
  • Bias detection happens after deployment
  • Cultural context often ignored in training
  • Global South perspectives underrepresented
Usage Ethics
Reactive approach addressing problems after AI is deployed
Creation Ethics
Proactive approach embedding values from the start
Ethics in Creation, Not Just in Use
Belem-LLM integrates moral reasoning, transparency, and cultural context directly into model training. Our approach transforms how AI systems learn and develop from their foundation.
Ethical Corpus Design
Balanced, cited, and inclusive training data that represents diverse perspectives and values.
Bi-Directional Modeling
Portuguese-English language pairs preserving cultural nuance and context
Bias Detection Layers
Embedded interpretability and bias-detection systems throughout the architecture
Open Academic Access
Shared resources for UNAMA and ANPAD research partners
From the Amazon to Appalachia
Academic & Social Impact
Academic Impact
  • Introduces "AI Literacy with Integrity" modules in business and marketing curricula
  • Trains faculty to evaluate and co-create ethical datasets
  • Develops new pedagogical frameworks for responsible AI education
  • Creates research opportunities for graduate students
Social Impact
  • Preserves Lusophone research heritage and cultural knowledge
  • Promotes equity in AI between Global South and North
  • Bridges technological divides across continents
  • Empowers underrepresented communities in AI development
Collaborating Across Continents
Belem-LLM brings together leading institutions from both hemispheres to create a truly global approach to ethical AI development.
Fulbright U.S. Scholar Program
Founding sponsor enabling international academic exchange and collaboration
Universidade da Amazônia (UNAMA)
Brazilian academic partner providing regional expertise and research infrastructure
ANPAD
Brazilian Academy of Management connecting business education with AI ethics
Pittsburgh Innovation Hub
U.S. technology and ethics research center driving responsible AI development

Supporting Partners (invited): OpenAI • NVIDIA • AACSB International • IBM Research Brazil • SteelCity.AI
Investment in Ethical Infrastructure
Initial Needs
  • GPU and cloud compute credits for model training
  • Corpus digitization and translation pipeline development
  • Research assistantships and fellowships
  • Cross-border workshops and knowledge exchange
Projected Outcomes
  • Open-access Belem-LLM 1.0 release
  • Ethics-by-Design curriculum toolkit
  • Dual-language academic journal integration
  • Sustainable research community
1
Stage 1
Infrastructure setup, corpus development, partnership formalization
2
Stage 2
Model training, curriculum piloting, faculty development programs
3
Stage 3
Public release, impact assessment, sustainability planning
Why Belem-LLM Matters Now
"AI mirrors its makers. We must train models — and minds — to see ethics as architecture, not afterthought."
Global Access
Ensuring AI development includes voices from the Global South and preserves linguistic diversity
Proactive Ethics
Building moral reasoning into AI systems from inception rather than retrofitting later
Educational Innovation
Creating new frameworks for teaching AI literacy with integrity across disciplines
The Belem-LLM Difference
What sets our approach apart is the integration of ethical considerations at every stage of development, from data collection to deployment.
Ethical Data
Curated corpora representing diverse perspectives and cultural contexts
Transparent Training
Open documentation of model decisions and bias mitigation strategies
Community Validation
Continuous feedback from academic and cultural stakeholders
By embedding ethical reasoning directly into the model architecture, Belem-LLM creates AI systems that don't just process language — they understand context, respect cultural nuance, and reflect human values.
This isn't just about building better AI. It's about building better patterns for how we create technology together.
Join the Movement
Partnership Opportunities
Research Partners
Collaborate on ethical AI development, contribute to corpus design, and co-author publications on responsible AI practices
Funding Partners
Support computing infrastructure, fellowships, and workshops while gaining early access to research findings and tools
Policy Partners
Help shape ethical AI frameworks and standards that can be adopted across institutions and industries globally
Building Better Patterns for a Better World
AI mirrors its makers. Belem-LLM trains models — and minds — to see ethics as architecture, not afterthought. Join us in shaping the next generation of human-centered AI.
2
Continents
Bridging Brazil and the United States
4+
Partner Institutions
Growing network of academic and industry collaborators
1
Shared Vision
Ethical AI that serves humanity
Contact: Dr. Mark Lennon — Fulbright U.S. Scholar, UNAMA Brazil - [email protected]
Where the Amazon Meets the Algorithm
AI App - The Number Trick
Spot the Bot: AI Hallucination Game
Humans > Algorithms
AI App - Career Advice
Contact Us