Where AI Fits – and Where to be Cautious
AI is transforming workflows across public health and it has it’s place in public health documentation. AI is a great tool to organize brainstorming and get basic information on a topic. However, when it’s time to apply for grants and cite your sources, you need to exercise caution when using AI. Grant application guidelines around AI are becoming more strict and the data you use in your application needs to be credible, able to be tracked back to it’s source and valid. RHIhub (Rural Health Information Hub) meets all those criteria.
This is part 3 in a blog series from the expert webinar hosted by Champ Software, featuring Erin Lord-Kunz from RHIhub. Read Parts 1 and 2 of the series:
Pulse Check Points
The Right Role for AI in the Public Health Grant Application
Sarah Dobson, and expert in helping researchers communicate the value of their work for funding purposes, has helped clients secure NIH funding. In the video, “Basic Rules for Leveraging AI to Write Research Grants” she cites the article “10 Simple Rules to Leverage Large Language Models for Getting Grants.” Watch the video and read the cited source for details. The short version is that there is a place and way to use AI and places and times to not use it.
AI is effective for:
- Drafting narratives
- Brainstorming ideas
- Structuring content
But it should not be treated as the final authority for:
- Citing statistics
- Validating data
- Supporting funding claims
Because fundamentally, these systems generate outputs based on patterns in language—not verified truth.
The Risk of Hallucinations and Outdated Data
AI may produce fabricated or misattributed statistics. This behavior is documented as “hallucinations” in research. In the article “10 Simple Rules to Leverage Large Language Models for Getting Grants” the authors share that “Generative AI models such as LLMs are known to “hallucinate”—or fabricate—facts and references in response to prompts, given the nature of their training.”
AI may also provide answers that appear current but rely on outdated or incomplete information, requiring users to validate sources independently. When writing prompts for AI be sure to include data specific information like, “find me the cost to rural health agencies to provide maternal child health services and use references that are from 2025 or more recent.”
The Growing Compliance Landscape
As scrutiny increases, so do the consequences.
Federal grant applicants should be cautious: NIH has publicly said it uses technology to detect AI-generated content in research applications, signaling a growing expectation that submissions reflect original thinking and properly validated source material.
For readers who want to follow someone closely connected to both local public health and AI, Aaron Davis, Director of the Center for Public Health Initiatives at Wichita State University, regularly shares insights on policy, implementation, and emerging issues. View Al Aaron Davis’ LinkedIn profile.
From Data Integrity to Documentation Integrity
Once you have trusted, verifiable data, the next challenge is ensuring that same level of integrity carries through the rest of your work.
In practice, this is where many public health workflows begin to break down.
Data may start out strong—sourced from credible, curated platforms like RHIhub—but as it moves into daily operations, it can become:
- Fragmented across spreadsheets and systems
- Inconsistently documented across programs
- Difficult to tie back to original goals or funding requirements
Maintaining data integrity throughout the lifecycle of a program requires more than good intentions—it requires structure.
That includes:
Consistent Documentation
Public health teams must document services, interventions, and interactions in a standardized way across staff and programs. Without consistency:
- Data becomes difficult to compare
- Reporting becomes manual and time-consuming
- Variability introduces risk in audits and evaluations
Consistency ensures that what is documented in the field aligns with what is reported at the program and funding level.
Structured Data Capture
Free text and disconnected notes limit the ability to analyze and report on outcomes.
Structured data capture—using defined fields, standardized forms, and consistent frameworks—allows teams to:
- Aggregate data across populations
- Track trends over time
- Align documentation with specific grant deliverables or health priorities
This is especially important when connecting individual-level services to broader initiatives like CHIP goals or SDOH measures.
Reportable, Defensible Outcomes
Ultimately, the purpose of both data and documentation is to demonstrate impact.
Public health departments are increasingly expected to show:
- What services were delivered
- Who was reached
- What changed as a result
That requires documentation that is:
- Complete
- Consistent
- Directly tied to measurable outcomes
Without that connection, even well-designed programs can struggle to demonstrate their effectiveness.
Maintaining the “Chain of Custody” Beyond Data Sources
The webinar emphasized the importance of a data “chain of custody”—knowing exactly where your data comes from.
That same concept applies internally.
Public health teams must be able to trace:
- A reported outcome → back to documented services
- Those services → back to program goals
- Those goals → back to validated data and identified needs
When that chain is intact, organizations can move confidently through:
- Grant applications
- Reporting requirements
- Program evaluations
From Insight to Impact
The future of public health depends on more than access to good data—it depends on the ability to carry that data through every stage of the workflow.
That means alignment between:
- Trusted data sources that provide accurate, verifiable information
- Responsible use of AI to support—not replace—critical thinking and validation
- Strong documentation systems that ensure data remains structured, consistent, and reportable
When these elements work together, public health teams can move from:
- Gathering information
➡️ to documenting action
➡️ to demonstrating measurable impact
Resources like RHIhub—and the experts behind them—play a critical role in helping teams start with trusted information.
Starting with good data is only part of the equation.
To fully demonstrate impact, teams need a documentation system for public health that allows them to:
- Track services consistently across programs
- Tie those services directly back to program goals and funding requirements
- Measure and report on outcomes at the individual, family, and community level
- And pull that data together in a structured, reportable format to support grant applications and ongoing funding
It is this combination of trusted data, thoughtful workflows, and purpose-built documentation systems that ultimately allows organizations to turn insight into action—and action into outcomes they can prove.
Key Takeaways
- AI is a powerful assistant—but not a source of truth.
It can help structure ideas and accelerate writing, but it cannot reliably validate statistics or replace authoritative, source-based data. - Hallucinations and outdated data are real risks in high-stakes work.
AI-generated content may sound accurate while relying on fabricated or obsolete information—making independent verification essential. - Compliance expectations are evolving quickly.
With organizations like NIH signaling the use of AI-detection tools, grant submissions increasingly require original thinking and clearly validated sources. - Data integrity must extend beyond sourcing into daily workflows.
It’s not enough to start with trusted data—public health teams must ensure that documentation, reporting, and outcomes maintain that same level of accuracy and consistency. - The ability to prove impact depends on connected systems.
Aligning trusted data, responsible AI use, and a purpose-built documentation system enables teams to track services, measure outcomes, and confidently support grant applications.