Smart Money Doesn’t Invest In Witless Enterprise AI For Healthcare
AI for healthcare has come a long way. Some commentators envision artificial intelligence threatening humanity in a nightmarish future. Others fear AI will take their jobs in the next year or two. Meanwhile, investors looking for opportunities in the sector are working hard to determine how AI will change how we live, work, and play.
The futurists are not entirely wrong. AI is going to impact critical life and death diagnosis and treatment decisions made by clinicians in the coming years. But that power will reside in a medical device that is keeping a patient alive in a hospital bed, not an evil robot worthy of Hollywood. Regulators, ethicists, and others are now developing guidelines on how AI technology should work so the new market can grow.
AI for healthcare. Yes, please.
The FDA’s recently proposed framework for AI software in medical devices accepted that the technology was necessary for healthcare because it improved its performance with real-world use and experience, adapting and optimizing in real-time, and continuously improving health care for patients. According to the FDA, AI could potentially transform American healthcare. I agree. AI is already changing how we look for a vaccine for the coronavirus, according to The Lancet.
In medicine, patients, families, and stakeholders will accept AI technology if it improves care. However, nobody will turn their care over entirely to a machine, no matter how much it has learned. On the contrary, they want AI that helps their doctors treat them, that works at the behest of their medical team, that expedites tests that they need to make decisions.
That sort of AI is likely to flourish and benefit humanity in the coming years, growing despite the rhetoric about job losses and the apocalypse. I believe the successful products will have three characteristics – they will be explainable to whoever uses it, verifiable to see exactly how it benefits them, and interactive, meaning users can adjust the technology to their needs.
When humans understand technology, see why it works, and interact with it, they create incredibly powerful virtuous cycles where data and discoveries lead to conclusions and insights that deliver more data and discoveries. Businesses that hone this process can achieve stupendous benefits, often developing so-called moats or barriers to competitors, securing sometimes exponentially more value.
Nobody wants AI alone to determine one’s course of care.
Investors, therefore, would be foolish to invest in AI that would aim to replace radiologists.
AI that tries to replace humans, rather than learning how it can be most useful to humans, is not good AI. Wise investors are focused on making AI people love to work with because that’s where real value and intelligence are unlocked. Once we start accessing that value and intelligence, the dividends will resolve any disputes about AI’s next steps.
When it comes to AI and medicine, there is only one path forward: physicians will retain control of explainable, transparent, interactive AI. The case is no different for investors, manufacturers, the Internet of Things, or whatever else the future holds.