The potential roles and risks of Artificial Intelligence in public health: A scoping review
This scoping review identifies 44 potential roles for artificial intelligence in public health, primarily involving data analysis and communication, while highlighting critical risks such as bias, ethical concerns, and the exacerbation of health inequalities, ultimately calling for rigorous evaluation, community engagement, and careful implementation to ensure equitable benefits.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine public health as a massive, busy city trying to keep its citizens healthy. The city planners (public health officials) have limited time, money, and energy, but they have a huge job: preventing disease, teaching healthy habits, and responding to emergencies.
Now, imagine Artificial Intelligence (AI) as a brand-new, super-fast, super-smart assistant that has just moved into City Hall. This assistant can read millions of books in a second, predict traffic jams before they happen, and write speeches in any language. But, like any new tool, it has a catch: if you don't use it carefully, it might accidentally make the city's problems worse.
This paper is a scoping review, which is like a "map-making" expedition. The authors didn't test the AI themselves; instead, they gathered 67 different reports (like news articles, expert opinions, and studies) published between 2020 and 2025 to answer one big question: "What could this AI assistant actually do for our city, and what could go wrong?"
Here is what they found, broken down into simple terms:
1. The Superpowers (Potential Roles)
The researchers found 44 different ways this AI assistant could help the city. They grouped these into eight main neighborhoods:
- The Town Crier (Communication): The AI can write messages tailored to specific groups. For example, it could explain a health warning in simple language for a child, or in a specific dialect for an elderly community. It can also spot rumors spreading online and help correct them quickly.
- The Teacher (Education): It can act as a personal tutor for public health workers, creating custom training modules that adjust to how fast a student learns.
- The Emergency Manager (Disasters): If a storm hits or a crisis occurs, the AI can help decide where to send ambulances, how to evacuate people, and how to translate instructions for different groups instantly.
- The Librarian (Evidence Reviews): Public health workers often have to read thousands of research papers to make decisions. The AI can act like a super-librarian, quickly finding the right papers, summarizing them, and even writing the final report.
- The Fortune Teller (Forecasting): The AI can look at past data to predict future trends, like how many people might get diabetes next year or when mosquito populations will spike.
- The Health Coach (Interventions): It can help design programs to get people moving or eating better, and even use chatbots to talk to people about their health habits.
- The City Manager (Organizations): It can handle boring paperwork, organize huge piles of data, and help leaders write grant applications.
- The Security Guard (Surveillance): It can use cameras to check if people are wearing masks or keeping their distance (though the paper notes this is a specific, limited use).
2. The Hidden Traps (Potential Risks)
Just like a powerful tool can be dangerous in the wrong hands, the paper warns that AI comes with serious risks. The most common dangers found in the reports were:
- The Biased Mirror: If the AI is trained on data that only represents some people (like wealthy neighborhoods), it will "learn" that those people are the only ones who matter. It might then give bad advice to minority groups or poor communities, making health inequalities worse instead of better.
- The Hallucinating Writer: Sometimes, especially with the new "generative" AI (like chatbots), the tool might make things up. It could write a health article that sounds perfect but is factually wrong, or create fake images.
- The Privacy Leak: To work, AI needs data. The paper worries that using this data could accidentally expose people's private health information.
- The Trust Breaker: If the public feels the AI is unfair, secretive, or makes mistakes, they might stop trusting public health officials entirely. If people don't trust the city, they won't listen to its advice.
- The Skill Erosion: If workers rely too much on the AI to do their thinking for them, they might forget how to do their jobs properly. If the AI fails, the humans might not know how to fix it.
- The Digital Divide: Not everyone has a smartphone or good internet. If public health relies on AI tools, the people who need help the most (those without tech access) might be left behind.
3. The Verdict
The authors conclude that while AI has the potential to be a superhero for public health, we are currently in the "blueprint" phase. We have a list of cool ideas, but we don't have enough proof yet that these ideas actually work or that they are safe.
They argue that before we let this AI assistant run the city, we need to:
- Test it rigorously: Prove it actually saves lives and money.
- Check for bias: Make sure it doesn't discriminate against vulnerable people.
- Build guardrails: Create specific rules for public health (not just for hospitals) to ensure it's used ethically.
- Listen to the people: Involve diverse communities in the design process so the tool works for everyone, not just a select few.
In short: AI is a powerful new engine for public health, but if we don't steer it carefully, it could crash the car. We need to drive slowly, check the map, and make sure everyone in the car is safe before we hit the gas.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.