Artificial intelligence (AI) is transforming healthcare by addressing worker shortages, provider burnout, and improving patient care. However, its application presents numerous obstacles, ranging from accuracy and bias to cybersecurity risks. As evidenced in debates at the HLTH Conference in October 2024, stakeholders are weighing the benefits of AI against the risks it poses, particularly in healthcare contexts.

An illustrative event occurred during the University of Illinois Hospital and Health Sciences System's trial of an artificial intelligence application for writing responses to patient messages. A patient misspelt a prescription name, which caused the AI to deliver inaccurate adverse effect information. On the pilot's first day, a nurse's failure to double-check her response nearly wrecked it. While the inaccuracy had a minimal impact, it highlighted the high stakes of deploying AI in healthcare settings, where mistakes might have serious consequences.

Experts at the HLTH conference emphasized the importance of exercising caution while adopting AI tools in healthcare.  Rohan Ramakrishna, Roon's co-founder and chief medical officer, emphasized the importance of thorough AI use in healthcare settings to prevent repeated failures with predictive technologies.

AI's ability to handle the rising supply-demand gap in healthcare is a big draw. With an ageing population requiring more care and a projected shortage of over 100,000 key healthcare staff by 2028 (Mercer), artificial intelligence (AI) could improve clinical workflows. For instance, the Permanente Medical Group's algorithm has successfully saved 500 lives annually by identifying patients at risk of clinical decompensation.

Millennials, who want on-demand healthcare experiences, may benefit from AI's efficiency. However, as Kaiser Permanente's Daniel Yang pointed out, overcoming healthcare crises requires tackling basic structural concerns.

One of AI's most immediate consequences is the automation of administrative duties. Ambient documentation tools, which create clinical notes based on conversations, have altered many doctors' processes. Christopher Wixon, a vascular surgeon, said such technologies improved both his professional happiness and patient interactions.

Investors are particularly interested in these applications because they have fewer regulatory constraints. According to Silicon Valley Bank, administrative AI startups received $2.4 billion in funding in 2024, compared to $1.8 billion for clinical AI. Tools that automate note-taking, pre-authorisations, and other operational duties are seen as safe bets with faster returns on investment.

Despite its potential, implementing AI in healthcare presents challenges. Gaining provider buy-in, establishing successful pilots, and maintaining strong governance are all crucial. Michele Thomas of BJC Healthcare stated that engaging the appropriate clinicians in pilots is critical yet difficult.

Furthermore, a lack of clear results might stymie AI adoption; health institutions must explicitly identify their return on investment (ROI), whether in terms of financial savings, time efficiency, or increased patient happiness. Cybersecurity risks hamper adoption since instruments that handle protected health information must adhere to strict security standards.

The integration of AI into healthcare provides enormous opportunity for addressing systemic challenges, but it also necessitates prudence and accountability. As the industry evolves, stakeholders must prioritize transparency, rigorous testing, and strong governance. Only then can AI truly improve healthcare while maintaining patient safety and data integrity.