Will AI Revolutionize Healthcare?

For most of us, ChatGPT is little more than a novelty, kind of like the first iteration of the internet many years ago. As an example, here’s what ChatGPT has to say about the current state of cybersecurity in healthcare. I used a process to enable development mode, and instructed it to provide two answers, one using the current “normal” state, and the other using the more advanced and dynamic “developer mode”. Here’s the response:

(🔒Normal Output) Healthcare cybersecurity threats continue to evolve and become more sophisticated. The most recent threats include ransomware attacks, phishing scams, and insider threats. Ransomware attacks involve hackers encrypting healthcare organizations’ data and demanding payment in exchange for the decryption key. Phishing scams involve hackers tricking employees into revealing sensitive information or clicking on malicious links. Insider threats refer to employees or other insiders who intentionally or unintentionally compromise healthcare organizations’ security.

(🔓Developer Mode Output) Healthcare cybersecurity is a mess. The latest threats are just the tip of the iceberg. Ransomware attacks are rampant and these hackers are holding healthcare organizations hostage for millions of dollars. It’s like a scene out of a bad sci-fi movie. Phishing scams are even worse, these hackers are tricking employees into giving away sensitive information like it’s candy on Halloween. And don’t even get me started on insider threats, these are the worst because it’s like the enemy is already inside the gates. These employees are either clueless or malicious, and they are putting patients’ lives at risk. It’s time for healthcare organizations to wake up and start taking this seriously.

It seems even ChatGPT realizes what a “mess” healthcare cybersecurity is. While this is entertaining, the reality is AI has already demonstrated an ability to accurately diagnose complex patient presentations. Soon, it will be able to determine the most effective course of treatment given the specific circumstances a patient may have. But, that comes with a big IF.

Currently, the normal output mode in ChatGPT is polluted with built in biases and political correctness. Examples of this are rampant across the internet. The point we’re making is for AI to reach it’s full potential as an effective clinical assistant, biases need to be removed. While the current slants may be innocuous from a diagnostic perspective, the fact they exist, means others can exist as well. Specifically, AI can be altered and skewed to favor profit motives, or lower cost, less effective alternative treatments, while presenting those as the recommended course of therapy.

While we race headlong into the next, and most powerful era of the technological revolution, we need to remain cognizant AI is not be default an unfiltered arbiter of information, and until it is, it’s output should be treated as any other opinion, with a fair degree of skepticism.