It seems that everywhere you turn these days people are talking about artificial intelligence (AI) and machine learning.
So when I saw one of the panels at SXSW this year was titled “The Future of Machine Learning: Worth the Hype?” I had to go see if it really answered the question. The replay is currently available at that link if you want to decide that for yourself.
For me, the most interesting takeaways came from Finale Doshi-Velez, an assistant professor of Computer Science at Harvard University, whose core research in machine learning, computational statistics, and data science is inspired by — and often applied to — the objective of accelerating scientific progress and practical impact in healthcare and other domains.
For many of us, our first introduction to machine learning had to do with computers learning to play chess. In the beginning, what the computer knew was what a human told it about chess. But today, the computers are teaching themselves to play and they learn by playing the game over and over.
But, when it comes to our health, we can’t play games with people’s lives. That’s why Doshi-Velez noted that we need to pay attention to what we care about, what our human values are, first before we leave decisions up to AI or even their human engineer who is just trying to make a deadline.
And that comes back around to the title of this post and another conversation I listened to at SXSW between Michael Dell and Clay Johnston, the inaugural dean of the Dell Medical School. Dell Medical School at The University of Texas at Austin is the first med school in decades to be built from the ground up at a Tier 1 research university.
It’s the Michael and Susan Dell Foundation that makes that possible, but Johnston said he did look to the company Dell founded as a model for how the democratization of technology could be applied to building the school.
“If you went to a company 10 years ago and said, ‘hey, what’s technology?’ They would say that’s the IT department,” Michael noted. “If you go to a company today and say, ‘what’s technology?’ — that’s actually how we make our products and services better. That’s how we interact with our customers — that’s everything that we do is about technology.”
He also talked about how within companies there is a cycle where you start with some data – maybe a little, maybe a lot – and you use that data to create a better product or service in some way. Each time taking on more data to help improve the next iteration of the product or service. Machine learning is that same cycle.
But the amount of data that can be collected today and the speed with which it can be analyzed is really rather mind-blowing.
What we do know is improvements are coming at a very fast pace…
While Doshi-Velez noted that it’s not just as easy as throwing all the data into a blender and out comes a cure for all cancers, the work our partner TGen is doing shows that it can help make a significant leap in the ability of doctors to treat children with cancer.
Johnston noted that today AI can do many things well, such as finding cancerous cells in pathology slides or reading chest x-rays to find tuberculosis. It can do so even better than doctors. So getting back to the title of this post, as the technology advances it necessitates a reimagining of how you train the doctor.
Where previously medical training was mostly about memorization, now healthcare professionals have quick access to more information that’s often more reliable than memory. That simple human and machine partnership that has already begun to take shape will only continue to expand.
While the thought of AI doctors might be scary, it’s really more of a matter of leveraging the AI for what it’s best at and giving the human doctors more time to focus on the truly human side of healthcare. Michael is optimistic that technology will assist the doctor for a greater result.
“What we do know is improvements are coming at a very fast pace,” he said. “It’s definitely a defining moment in terms of the future of all industries and society.”
As the potential patient, it’s a lot to try to wrap our heads around. But if you want to try to get a better understanding of machine learning without having to get a computer science degree, here’s a great explainer video I recently came across. Because of the bots.