This week, Moussa Doumbia announced that he has been chosen as one of the first AWS Machine Learning University Faculty Fellows for 2026-2027.
According to his website, Doumbia is a researcher with a Ph.D. in Mathematics from Howard University (2017) and a B.A. in Mathematics from the University of the District of Columbia (2007).
He has advanced data science and mathematics education, designing and teaching an introductory Data Science course, helping launch Howard University’s mathematics major with a data science concentration, and leading the proposal and development of all new courses in the new Bachelor of Science in Data Science program at Howard.
One of the main challenges in AI education is making sure teachers are prepared to teach it well.
In this week’s social media post, Dr. Doumbia explained that as one of the first AWS Machine Learning University Faculty Fellows for 2026-2027, he will teach artificial intelligence and machine learning and help other university faculty bring these topics into their classes and research.
He said this is important because one of the main challenges in AI education is making sure teachers are prepared to teach it well.
Workshops for the Next Generation of Innovators
In related news, Tyrone Taborn, CEO of Career Communications Group and creator of STEM City USA, recently announced that teachers and youth instructors can now become certified through Career Communications Group.
The certification includes youth workbooks and a teacher’s guide to help understand AI and new technologies.
The program is partly funded by the Maryland Department of Labor.
The public is also invited to attend the Career Communications Group AI NEXTGEN America Conference on October 9-10 at the Baltimore Convention Center for hands-on training.
Also, this fall, Howard University will introduce the Fundamentals of Artificial Intelligence Certificate, a certificate course that combines foundational AI concepts, an introduction to AI tools and techniques, ethical and responsible AI practices, and the application of AI concepts across various disciplines through specialized coursework.
This follows the June 2024 announcement of Howard University’s AI initiative.
The initiative is designed to address ethics, equity, and societal impact; research and innovation; education and skill development; and operational efficiency.
The initiative also established the President’s Artificial Intelligence Advisory Council. The group provides direction and implementation for AI across the institution, including academic programs, university offices and units, and strategic planning activities.
Building Skills and Opportunity
In 2025, the College of Engineering and Architecture (CEA) started an “AI Tinkery” workshop series.
The workshops are led by Howard Prioleau, a PhD candidate in computer science who specializes in artificial intelligence, machine learning, and natural language processing.
According to Howard University’s campus publication, the AI Tinkery course helps future engineers, architects, and tech leaders gain the skills they need for the digital world.
Each semester this past year, Dr. Legand Burge, a computer science professor and member of the President’s AI Advisory Council, and Prioleau have hosted a workshop.
The first workshop in fall 2025 focused on how generative AI systems use large datasets to recognize patterns and create text, images, and audio.
Participants also learned about writing clear prompts, considering the task, context, audience, constraints, and desired results.
The spring 2026 workshop taught how to structure prompts with guidelines for tone, format, and scope.
Participants explored using multiple perspectives in queries and finished by learning how to build prompt libraries for more strategic use.
Prompt engineering is now regarded as an essential skill that many professionals will need to develop.
Prompt engineering involves both coding and communication abilities. Prompt engineers play a key role in teaching non-technical users how to interact with generative text systems.
Individuals with IT service or editorial backgrounds can combine their expertise with prompt engineering skills to excel in this emerging field.
Mastering AI Prompts
A recent three-hour seminar by Margot van Laar, technical staff at Anthropic, focused on prompt engineering for large language models (LLMs).
Margot emphasized that prompt writing is a foundational skill for AI engineers and outlined methods to improve, adapt, and create prompts.
Using real customer examples from Claude, she highlighted common issues such as outdated prompts influenced by legacy policies and processes.
Margot recommended regularly testing and validating prompt changes to ensure effectiveness.
The seminar addressed three scenarios: standard control cases, edge cases in which detailed prompts prevent model failures, and situations in which human intervention is necessary.
Through a case study with Meridian’s support bot, she identified gaps in controls, capabilities, and policies, including challenges such as prepaid disclaimers.
Tools like VZero help users identify prompt failures, while debugging supports problem resolution.
Margot advised: if a prompt’s guidelines, policies, and data are unclear to you, they will likely be unclear to the model as well.
If a prompt’s guidelines, policies, and data are unclear to you, they will likely be unclear to the model as well.
Alexander J., founder and CEO of ElysiumNetwork, raised a critical question on social media: What if every prompt, code snippet, and design shared with an AI tool is stored for years and later used to train models you may have to pay to access?
He noted this is not hypothetical but the current default.
According to Anthropic, consumer users must opt out by a deadline, or their chats and coding sessions will be retained for up to five years for model training; business tiers are exempt.
OpenAI retains chats until they are deleted, then schedules their removal within 30 days, but may keep de-identified data indefinitely for “security or legal obligations.”
API data is kept for 30 days, with enterprises able to request zero retention.
Google states that Gemini can train on prompts, files, photos, and screen captures, and even after opting out, data may remain for up to 72 hours.
Your intellectual property is not stolen, but is legally absorbed, transformed, and resold. Consider: what would you build differently if you assumed this as the norm?
