Let's talk about AI, again... Hype vs Reality!
· Dr. Ramy Azzam
I recently finished an intense course on "AI in Medicine" and it got me contemplating... so of course I jotted it down.
The term "Artificial Intelligence", or what we all know to be "AI", dates back to the 1950s. John McCarthy, often referred to as the father of AI, coined the term in 1956 for the Dartmouth Conference, the first academic conference on the subject. Originally, AI referred to the creation of intelligent machines that could simulate human thinking and problem-solving. Fast forward, I am not sure if he will be happy with how the term is being used nowadays.
It is funny how things work out though, especially in medicine. Here we are, clinicians, cramming our brains with facts and figures, trying to become walking encyclopedias, so we can recall information as fast as possible when needed. I always felt a bit out of step with that approach. I remember this one time during a charity drive in med school - one of our top students, brilliant guy, completely froze when asked to do some basic math. He literally blurted out "Oh yeah, I don't do numbers". This was a lightbulb moment for me. Don't get me wrong, knowing your stuff is crucial but I couldn't shake the feeling that there must be and there is probably more than just ingesting facts and regurgitating them.
This epiphany lingered with me until Data Science became the avenue to pursue. To me, data science and its medical cousin, digital health seemed to offer a more analytical approach to healthcare. Beginning in 2017 with forays into R and Python, I observed AI's evolution from a niche academic pursuit to a driving force in medical innovation.
Fast forward to today, and we find ourselves in the midst of an AI revolution—or at least, that's what the headlines would have us believe. Generative AI models like the GPTs and the Claudes have captured the public imagination, leading to a surge of mainstream interest and investment in "AI" technologies.
This surge in popularity has brought with it a shift in terminology and expectations. Terms like AGI (Artificial General Intelligence) and ASI (Artificial Superintelligence) have entered the public discourse, often used by those with limited understanding of the underlying technology. Marketing teams have seized upon the AI hype, slapping the label on products and services that may have only tangential connections to true AI capabilities.
As academics and technical professionals, we find ourselves at a crossroads. Should we embrace this mainstream hype and adopt the new terminology, or should we stand our ground and adhere to the original, more rigorous definition of AI?
I recently watched a video by Linus Sebastian (https://youtu.be/nGIpdiQrFDU?si=-zGfW1l2SJh2ufgL) that resonated with me on this topic. Linus eloquently breaks down the difference between narrow AI (what we currently have) and AGI (what many people mistakenly believe we have). He emphasizes the importance of understanding these distinctions, especially when it comes to critical applications like healthcare and autonomous vehicles.
I find myself agreeing with much of what Linus says. While the current advances in AI are undoubtedly impressive, they are still far from the science fiction vision of truly intelligent machines. What we have today are highly sophisticated pattern recognition and prediction systems, not systems capable of general intelligence.
As professionals in this field, I believe we have a responsibility to educate and clarify. We should embrace the excitement around AI while also tempering expectations and correcting misconceptions. This means:
Being precise in our language, distinguishing between narrow AI and hypothetical AGI.
Highlighting both the capabilities and limitations of current AI systems.
Encouraging critical thinking about AI claims, especially those made in marketing materials.
Continuing to push the boundaries of what's possible while maintaining scientific integrity.
The journey from early machine learning experiments to today's sophisticated AI applications has been remarkable. As we look to the future, it's imperative that our excitement about AI's potential doesn't overshadow the need for accuracy and ethical consideration. By maintaining this balance, we can guide AI's development in healthcare responsibly, ensuring it genuinely benefits everyone.