AI in Pandemic Prediction: Navigating Benefits, Risks, and Guardrails

JJohn August 14, 2023 3:28 PM

Artificial Intelligence (AI) holds remarkable potential in predicting and detecting pandemics early. However, it could also lead to misinformation spread if not properly regulated. This article examines the dual-edged nature of AI in the context of pandemic prediction, emphasizing need for guardrails, with insights from BlueDot's founder, an AI-based company that detected COVID-19 early.

Balancing AI's potential with responsible use

The power of artificial intelligence (AI) was demonstrated when it was used for early prediction and detection of the recent pandemic. However, this technology also has the potential to spread misinformation if it's not carefully controlled. As the founder of BlueDot, one of the first companies to detect COVID-19, pointed out, it's crucial to ensure that in the rush to leverage AI's capabilities, we don't inadvertently cause harm by contributing to the spread of misinformation.

The misinformation risk of large language models

Large language models (LLMs), a form of AI that uses vast sets of data to predict and generate text, are susceptible to 'hallucination,' or generating false information. This potential for AI to fabricate information underlines the importance of establishing guardrails around its use. Without these protective measures, there's a risk that LLMs could unwittingly contribute to the spread of misinformation, which would be counterproductive and potentially harmful in a pandemic situation.

AI's triumph: BlueDot's early COVID-19 detection

BlueDot, a Toronto-based company, gained recognition for being one of the first to detect the signs of SARS-CoV-2, the virus causing COVID-19. Utilizing AI's ability to process large quantities of information, BlueDot analyzed tens of thousands of articles daily in numerous languages. This vast data analysis led to the early detection of an article reporting a 'pneumonia of unknown cause', thereby indicating the emergence of what would later be identified as COVID-19.

AI's capabilities extend beyond data processing to include the rapid identification of emerging diseases and sending early warning signals. These benefits were part of the founding motivation behind BlueDot, which sought to respond to infectious disease emergencies quickly and precisely in ways not possible in the academic sphere. Yet, attempts to leverage AI are anchored in trust, and there's been significant erosion of trust in recent years, posing a challenge to AI's broader acceptance and deployment.

While AI's capabilities are impressive, it cannot be the only tool employed to improve healthcare system readiness and resilience. AI models work by detecting relationships between words, not necessarily facts. As such, not all misinformation will be detected by AI. A possible solution to this challenge could involve experts from various fields helping to determine the truth and improve AI models' ability to spot misinformation. An integrated mix of expertise in epidemiology, public health sciences, data science, machine learning, and engineering is needed to build and enhance this technological infrastructure.

More articles

Also read

Here are some interesting articles on other sites from our network.