Trust at our core

Our principles for generative AI in healthcare

As pioneers in healthcare technology, we recognize the crucial role we play in incorporating artificial intelligence into healthcare workflows. We firmly support and align to leading AI ethical and safety principles.

BastionGPT's Healthcare Generative AI Principles

We have established a complimentary, and more focused set of principles, specific to healthcare generative AI, that act as our north star to guide our innovative journey to elevate patient care and reduce healthcare workforce burnout, focusing unwaveringly on the safety, privacy, and well-being of patients.

These principles aim to foster trust and transparency with our diverse community and reflect our steadfast commitment to maintaining the highest standards of integrity, accuracy, and reliability in our services. We encourage other organizations to embrace these guidelines, helping to realize a safe and transformative integration of technology and healthcare, with medical professionals at the forefront, ensuring a human-centric and ethical approach to patient care.

Principle 5: Guard Patient Interaction
“Generative AI must not directly provide medical advice to patients unless monitored and strictly validated.”

Generative AI is advancing rapidly, creating boundless possibilities. However, when it comes to providing medical advice, these solutions should not directly communicate with patients unless the outputs are strictly controlled and thoroughly vetted by both industry experts and the relevant regulatory bodies. The distinct risk of AI hallucinations and biases in healthcare topics necessitates medical professionals, who are trained and experienced, to supervise interactions. Medical professionals are essential in ensuring the accuracy and reliability of the advice given, recognizing the limitations of AI, and preventing potential harm to patients.

Principle 4: Uphold Privacy and Security
“All personally identifiable information must be kept private and secure.”

The influx of insecure AI services, often cheaper and below standard, poses significant risks to the security and privacy of sensitive personal information. The integrity and confidentiality of patient information must never be compromised. All AI healthcare services must align with healthcare standards and comply with regulatory requirements to avoid breaches that could jeopardize patient wellbeing. Establishing robust privacy controls and security measures are critical in maintaining trust and safeguarding patient interests.

Principle 3: Prioritize Evidence-Based Medicine
“Source data for generative AI must prioritize evidence-based medicine and reputable sources.”

Misinformation and biases are prevalent, even in reputable sources across the internet. Generative AI in healthcare must be anchored in evidence-based medicine, relying on trusted, scientific sources to inform its knowledge base. A stringent, proactive approach is essential to mitigate the risk of circulating harmful and inaccurate data points, improving the accuracy, reliability, and safety of the AI-generated outputs.

Principle 2: Promote Transparency and Caution
“Generative AI must transparently communicate its propensity for errors even when used by medical professionals.”

The incredible capabilities of generative AI can lead to a false sense of security and overshadow its inherent limitations and potential for mistakes. It is crucial to maintain transparency about AI’s limitations and encourage both users to approach the AI outputs with caution and critical evaluation. The emphasis should be on ensuring the responsible use of AI, with medical professionals maintaining vigilance and scrutiny to detect any inaccuracies, thereby preventing potential harm to patients.

Principle 1: Maintain Human Oversight
“A medical professional must mediate the provision of medical advice or information from AI to patients.”

The interface between AI and patients, especially in delivering medical advice or information, must be sufficiently supervised by medical professionals unless appropriate safeguards are robustly implemented. These safeguards can include strict governance around AI outputs, ongoing monitoring by healthcare professionals, and rigorous validation of medical accuracy. We must keep healthcare human-centric, with clinicians retaining control to ensure the safety, accuracy, and appropriateness of the advice provided.

“Every decision we make echoes in the well-being of a patient. Our commitment is not just to innovation, but to safe and ethically-driven innovation that always places human health and trust at its core.”
Josh Spencer
BastionGPT Founder