Augmented Human Intelligence in the Age of AI Workshop

Abstract: Po-Shen Loh

AI Is Here. How Can We Help All the Humans?

Researchers and companies are racing to be the first to create Artificial General Intelligence, which explicitly intends to broadly surpass human intelligence...

What then, will become of humans? It is now urgent to rapidly uplift human intelligence, as well as to unite humanity in a collaborative spirit, for the survival of the human race.

The speaker brings a unique background that bridges all of these worlds. He is on a mission to build a more thoughtful human world. He served as the national coach of the USA International Math Olympiad team for a decade, but also regularly interacts with students from underprivileged schools with low exam scores. He is also a math professor at Carnegie Mellon University, and a social entrepreneur and serial inventor, with a firsthand understanding of business, technology, employment, and scale. His innovations combine fields from math to the performing arts.

In this talk, he will share some of his recent work, which uses the perspective of game theory to align human incentives in a novel way. He invented new win-win ecosystems that facilitate mass-scale education of problem-solving (how to invent new solutions to previously-unseen problem types, which is not regularly taught in schools), while concurrently creating a social fabric of trusted and thoroughly human relationships that incentivize high-skill people to contribute their skills to others, rather than selfishly accumulating resources. He will also share his personal journey which helped him accumulate these insights about human incentives: by embarking on the highest-density public speaking tour ever conducted by a professor.

This work has previously been covered in CNN and the Wall Street Journal.

About the KeyNote speaker: Carnegie Mellon professor Po-Shen Loh is on a mission to help humanity thrive in the age of AI. He crosses over between disciplines from math to technology to performing arts, while perpetually on tour, speaking to and learning from audiences around the globe. He uses these experiences to innovate win-win (highly scalable) ecosystems to solve real world problems, ranging from reinventing learning to rethinking the paradigm for pandemic control.

Abstract & Bio: Ernest Fokoue

Artificial Intelligence as an Invitation to Transformation and Evolution: Embracing Humility, Harnessing Potential, and Rediscovering Ourselves

Artificial Intelligence—particularly Large Language Models (LLMs)—is not just a tool; it is an invitation to engage with the world in profoundly new ways. A golden opportunity to stretch and expand the boundaries of our long held and cherished paradigms. Indeed, by producing human-like text, code, art, and reasoning, AI challenges us to reconsider what it means to create, learn, be conscious, and even know ourselves. This talk explores how AI’s successes (and cautionary tales) invite us to:

Practice epistemic humility — recognizing the novelty of AI’s capabilities while acknowledging its limitations.
Harness AI’s latent power — unlocking new vistas in science, art, and philosophy.
Live more examined lives — leveraging AI as a mirror to Socrates’ dictum: "Know thyself."

From accelerating drug discovery to democratizing education—while navigating risks like bias and misinformation—AI offers a golden opportunity for growth. But only if we approach it with curiosity, wisdom, and humility.

About the Presenter:

Dr. Ernest Fokoue developed an early passion for philosophy, mathematics, and metaphysics, inspired by his parents' dedication to learning despite their struggles. With a rich academic background, including a Ph.D. in Statistics, an M.Sc. in Neural Computation, and a B.Sc. in Mathematics, Ernest is a staunch advocate for intellectual unity, transcending academic boundaries.

As a professor of statistical science, his mission is to empower students with state-of-the-art knowledge in statistical machine learning and data science, guiding them to become world-class experts in their fields. Passionate about the power and beauty of mathematics, Ernest remains committed to sharing his expertise generously, inspired by the belief that "As a man thinketh in his heart, so is he." His love for learning and his devotion to teaching are matched only by his pride in his daughter Ellie, who embraces both the tangible and ineffable in her art and fashion design.

Tutorial: Independent Health Team

MASTERING LOCAL LLMs: FROM INSTALLATION TO FINE-TUNING

WHAT YOU'LL LEARN:

Transform your AI applications with privacy-preserving local large language models
Install & Configure Ollama with various models including Gemma, Llama, and Mistral
Build RAG Systems that leverage your own data without cloud dependencies
Optimize Parameters to reduce hallucinations and ensure consistent outputs
Implement LoRA Fine-tuning to customize models with minimal computing resources

REAL-WORLD APPLICATIONS:

Access vast knowledge, generate summaries, create personalized content, develop AI-powered assistance systems, support research with private analysis, and handle confidential information—all while maintaining complete data privacy and control.

HANDS-ON EXPERIENCE:

Run complete workflows on their own machines
Experiment with parameter optimization
Build a functional RAG system using medical data
Implement LoRA fine-tuning for specialized tasks

WHY ATTEND?

Privacy First: Learn AI techniques that keep sensitive data on your device
Resource Efficient: Run powerful models on consumer hardware
Customizable: Adapt models to your specific domain needs
Future-Proof Skills: Master techniques at the forefront of AI development

PREREQUISITES:

Basic Python knowledge, laptop with 16GB+ RAM

Tutorial: Daniel McSkimming

Using LLMs Locally with Ollama

Large language models (LLMs) have proven useful in a variety of tasks, including information extraction, summarization, translation, and more...

In this tutorial we will explore downloading and using LLMs locally with Ollama, dramatically reducing cost and restricting all prompts and data to your local computer, with or without specialized hardware. We will use an open-source browser extension to provide a graphical interface to LLMs, examine the available models and their uses, build and submit simple prompts for response, and describe the limitations of using the Ollama system.

About the Presenter:

Daniel McSkimming is the Director of Bioinformatics and Supervisory Data Scientist in the Veteran Health Administration’s National Precision Oncology Program, where he helps ensure veterans diagnosed with cancer receive the best care possible using precision medicine. [...] More recently, he has been piloting the use of large language models (LLMs) to aid oncologists in summarizing complex medical reports and adding structure to unstructured electronic health record (EHR) notes.

Abstract: Mounika Goruganthu & Sesha Vermuti

AdsorbCast : An AI platform that reads pressure and temperature data to predict and enhance the efficiency of adsorption processes...

Presenters: Mounika Goruganthu (Independent Health) and Sesha Vermuti (Linde inc.)

This presentation introduces AdsorbCast, an AI platform designed to optimize adsorption processes critical in various industrial applications. By analyzing real-time pressure and temperature data, AdsorbCast predicts process efficiency and provides actionable insights. This allows for immediate adjustments, leading to enhanced performance, reduced waste, and improved operational outcomes. We will discuss the underlying AI models, the platform's architecture, and case studies demonstrating its effectiveness.

Abstract: Joaquin Carbonara

Everything You’ve Always Wanted to Do—Now Possible with Generative AI

The first LLM widely available to the public, chatGPT was released Nov 30, 2022. Within just five days of its release, it reached 1 million users. This rapid adoption continued, and by January 2023, ChatGPT had surpassed 100 million users, making it the fastest-growing consumer application in history.

Lists of notable LLMs, such as those on Wikipedia and industry blogs, typically enumerate between 30 and 50 prominent models, including both proprietary (e.g., OpenAI's GPT-4.5, Anthropic's Claude 3, Google's Gemini, xAI's Grok, Meta's Llama 4) and open-source models with commercial licenses (e.g., Mistral, Falcon, DBRX, Gemma, Command R, Phi-3, XGen, Nemotron-4).

In this brief tutorial we'll demonstrate several application that may not be obvious to the general audience:

Write a Multiple Choice test or survey with Google Forms from a pdf in minutes
Fine tune a model with LangChain so that it only answers questions from the given source
Use Flask to create an interactive web page to show to the public any project (we'll choose an easy project to illustrate the concept).

PREREQUISITES:

python and Visual Studio Code.

About the Presenter:

With a Ph.D. in Mathematics (University of California in San Diego) and an M.S. in Computer Science (University at Buffalo), Professor Carbonara brings a wealth of knowledge to the intersection of these fields. His research delves into Data Science and Analytics, Computational Mathematics, and the application of technology in education, including recent work on Machine Learning and Large Language Models. His leadership experience includes founding the Professional Science Masters in Computational and Applied Mathematics (2010) and in Data Science and Analytics (2020) at Buffalo State, and he has published extensively and organized several conferences focused on data science and analytics.

Abstract: Zhen Liu

AI for personal finance

Almost everyone living in the modern economy needs to deal with financial or investment decisions. Hiring a professional to guide us is often not cost effective given our limited budget. Can we use LLM, which is getting better at texts and calculations day by day, to manage the gap before it gets out of control?

We offer a road map and many examples on decisions such as money management, tax planning, investment for retirement, and risk management.

About the Presenter:

Dr. Zhen Liu is an adjunct instructor at the University at Buffalo and the Buffalo State University. He has published in microeconomic theory, taught extensively in Economics, Finance, data science, and taught himself Python. He also served as a volunteer in many community organizations. Using AI to empower people is one of many things he enjoys doing.

Abstract: Angela Thering & Naila Ansari Catilo

Unlock the Power of Gen-AI: Essential Prompt Engineering Strategies for Instructors & Business Professionals

This upcoming workshop is designed to help educators and professionals use Generative AI more effectively. The session will introduce prompt engineering as the skill of writing clear, structured instructions that guide AI tools like ChatGPT to produce useful, relevant responses.

Presenters will highlight real-world applications—such as drafting client communications, redesigning lessons, summarizing readings, and crafting professional messages—and address common resistance to Gen-AI, including concerns about ethics, bias, and human creativity. A core message will emphasize keeping the “human in the loop”: users must verify, edit, and contextualize AI output to ensure accuracy and appropriateness.

The workshop will also introduce the RICE framework (Role, Instruction, Context, Expectation) along with four key strategies—Zero-Shot, Few-Shot, Chain-of-Thought, and Self-Consistency Prompting—to help participants get better results. Attendees will be encouraged to try out a strategy, reflect on their experience, and continue learning how large language models work to stay ahead of rapid AI developments.

About Angela Thering, PhD – Workshop Facilitator:

Angela Thering is a lecturer, instructional designer, and adult educator with a passion for making emerging technologies accessible, inclusive, and practical for educators and professionals. With 20 years of experience in Higher Education, she specializes in instructional innovation, faculty development, digital accessibility, and the integration of AI tools in teaching and learning.

Angela teaches in the Department of Higher Education Administration and Adult Education at Buffalo State University. She was part of the planning team for Augmented Human Intelligence in the Age of AI, a campus-wide faculty development event focused on exploring practical, ethical, and creative applications of Generative AI in higher education.

Angela also facilitates Buffalo State’s AI Book Club and AI Working Group and is a DS4EDU Fellow at Carnegie Mellon University, where she collaborates with experts in AI, data science, and learning analytics to explore best practices for technology-enhanced education. She works closely with faculty and staff to empower them to use tools like ChatGPT to streamline workflows, support students, and enhance creativity. She is also a New York State certified teacher in Social Studies, Teaching Students with Disabilities, and Career and Technical Education.

Angela is dedicated to demystifying the exploration of Generative AI and designing approachable, low-barrier learning experiences that support adults in confidently learning more about AI.

Abstract: Brian Barrey & Navya Mehta

Auto-Grader for Calculus Class: MathGraderAI Tutorial

Manually grading Calculus assignments can eat up to five hours of your day—hours better spent teaching and supporting students. MathGraderAI changes the game: our AI-driven grading platform automatically evaluates student work against a clear rubric, delivering consistent and fair scores every time. Beyond checking answers, MathGraderAI analyzes each student’s steps, logic, and methodology using state-of-the-art language models and algorithms.

For Educators:

Slash grading time, eliminate repetitive tasks, and access powerful analytics that reveal class-wide learning trends.

For Students:

Receive instant, personalized feedback that pinpoints errors, explains solutions step by step, and links to curated resources for deeper understanding.

Join our hands-on tutorial to see how you can customize MathGraderAI for your own classroom!

What to Expect

API Integration: Learn to integrate the ChatGPT API and Ollama models into your grading workflow.
RAG Pipelines: Build efficient pipelines that prep course materials for AI processing with Retrieval-Augmented Generation techniques.
Prompt Engineering: Discover best practices for crafting prompts that produce reliable, high-quality grading results.

Speaker Bios

Navya Mehta

Navya holds a Bachelor’s in Electronics and Computer Engineering and began her career as a software engineer. She is currently completing a Master’s in Data Science & Analytics and, after graduation, will join T‑Mobile as a Business Analyst Intern. Passionate about data storytelling and AI advancements, Navya spends her free time reading self‑help books, hiking, and experimenting in the kitchen.

Brian Barrey

Brian earned degrees in Mathematics and Data Science from Buffalo State University. He now teaches AP Calculus at Hutch Tech High School in the Buffalo School District, where he inspires students with real‑world applications of mathematical principles.

Augmented Human Intelligence in the Age of AI

Expand Your Understanding of AI

Keynote Speaker

Po-Shen Loh (Carnegie Mellon University)

Special Presentation

Ernest Fokoue (RIT)

Workshop Agenda

2:00 PM - 4:00 PM | Hands-on Concurrent Workshops

Room 1 -- CR202: Advanced Techniques Sampler

Room 2- CR204: Introduction to Gen-AI for High Quality Results

4:00 PM - 6:00 PM | Networking & Presentations (Main Floor - CR109)

Organizers & Supporters

Organized by:

Supported by: