Exploring AI and Prompt Engineering in Sports Science: Insights from Professor Greg Haff
- Jo Clubb
- Oct 15
- 5 min read
Artificial intelligence (AI) is reshaping the way we learn, work, and create. While its rapid development can feel overwhelming, for those in sport science and strength and conditioning, it presents a major opportunity.
In this interview, I sat down with Professor Greg Haff, one of the world’s leading figures in strength and conditioning research and education, to explore how AI and prompt engineering can extend the capacity of sports scientists, educators, and coaches.
Professor Haff is widely known for his research on resistance training, periodisation, and athlete monitoring. But in recent years, he has turned his curiosity towards AI - undertaking courses in prompt engineering, AI agents, and automation tools to explore how these technologies can be used in high performance sport and higher education.
What Is Prompt Engineering?
Prompt engineering is the practice of designing effective instructions for AI systems, such as ChatGPT, Claude, or Gemini. Rather than treating AI like a search engine, Haff describes it as “programming with language”.
He draws a parallel to coaching communication:
“If I’m coaching an athlete, I have to give them good contextual communication to get them to do something... and really prompt engineering is taking this tool, which is a large language model, and giving it instructions to help it become more precise."
A framework Haff finds particularly effective comes from the Google Prompting Essentials course, which suggests users “thoughtfully create really excellent inputs”. This acronym translates to the following steps: Task – Context – References – Evaluate – Iterate.
Iteration, he stresses, is the most important yet most overlooked step. AI outputs should never be accepted at face value; each interaction should be refined, checked, and improved.
Ethics and Education: Teaching AI the Right Way
AI is now embedded in higher education, but its use raises important ethical questions. Rather than banning AI in coursework, Haff designs learning activities that teach students how to use it responsibly.
For example, he allows students to use AI to generate a sample resistance training programme, but then requires them to evaluate that programme against recognised standards from the UKSCA, NSCA, or ASCA. This approach forces students to think critically about quality, source credibility, and alignment with best practice.
He also warns of the risk of “deskilling”, where students outsource thinking to AI and lose the ability to problem solve independently. The key, he argues, is to use AI as a learning amplifier, not a substitute.
He explains:
“People have to learn how to use it ethically. We're not going to stop people from using it... So we've got to teach them how to use it and use these kind of prompt engineering skills, I think [they] are going to help us educate the future sports scientist.”
Creating AI Agents
Once you’ve mastered prompting, the next step is AI agents: customised assistants that perform recurring tasks.
Haff uses ChatGPT’s GPT builder, Claude from Anthropic Academy, and Google’s Gemini to design agents for teaching and research. For example, he created a GPT that automatically generates individual learning plans for his students based on feedback he provides.
In a high-performance setting, similar systems could automate time-consuming tasks such as generating reports for different staff groups. One dataset could produce multiple tailored reports for coaches, athletic trainers, and nutritionists simply by adjusting the prompt language for each audience.
These skills, he notes, can be learned through resources like Vanderbilt University’s Prompt Engineering for ChatGPT and Google’s AI Agent Handbook.
AI in High Performance Sport
Looking to the future, Haff believes AI will have its earliest and most significant impact in athlete monitoring and analysis.
“We have all worked in high performance where we know that teams collect buckets of data and do very little with it, because they can't process it fast enough, in a timely way to actually action it... I think that's where things are going to be really accelerating.”
He sees AI being used to automate coding and performance analysis, integrate multiple data sources, and support personalisation at scale, delivering precision programming to large squads without overwhelming staff capacity.
In the longer term, he predicts new roles will emerge, such as AI Performance Strategists, Data Curators, and AI Ethics Officers, all working to ensure data-driven innovation remains ethical and human-centred.
Getting Started with AI
For practitioners looking to begin, Haff suggests three practical steps:
Experiment with a tool.
Start with a free AI model like Gemini, ChatGPT, or Claude, and dedicate little and often time each week to exploring prompts in your daily workflow.
Take a foundational course.
Recommended options include:
Learn ethically.
Avoid uploading athlete data or copyrighted materials, and be clear on what constitutes open-access information. AI is a tool to amplify expertise, not replace it.
For ongoing insights, Haff also recommends following Ethan Mollick’s Substack Useful Thing and exploring educational resources like Google’s Toward AI-Augmented Textbooks.
Final Thoughts: AI as a Tool in the Coaching Toolbox
In closing, Greg emphasises that AI should be viewed as one more tool in a coach or scientist’s toolbox - powerful, but not a replacement for human expertise.
“We still need the human expert. It's not like we're being replaced.”
He reminds us that strength and conditioning has always evolved with technology - from timing gates to force plates to GPS - and AI is simply the next progression. As long as practitioners remain curious, critical, and ethical, AI can extend both capability and impact across the performance environment.
Further Resources
Courses and Guides
YouTube Channels
Reading and Updates
Ethan Mollick’s Substack: Useful Thing
FAQs
1. What is prompt engineering?
Prompt engineering is the process of crafting effective instructions for AI systems to generate more accurate, context-aware responses.
2. Which AI tools are most useful for sports scientists?
ChatGPT, Claude, and Gemini each have different strengths. ChatGPT excels in coding and structured workflows, Claude in writing and reasoning, and Gemini in integration with Google Workspace.
3. Is AI safe to use with athlete data?
Only within secure, enterprise environments. Public models should not be used for identifiable or sensitive data.
4. Where should I begin if I’m new to AI?
Start with a free model and one short online course. Spend 20 minutes a week experimenting—just like developing a new skill in Excel or Power BI.
Watch the full conversation with Professor Greg Haff on the Global Performance Insights YouTube channel to explore these ideas in greater depth.