Power And Prediction
As the artificial intelligence (AI) frenzy reaches new heights, everyone is focused on understanding possibilities. Much will be written on the topic, with some exploiting the frenzy and others offering valuable insights. A book I’m reading titled Power and Prediction fits the category of valuable insights. As I read it, multiple thoughts are swirling. The book looked at historical progressions of major general-purpose technologies (steam, electricity, Internet, etc.) that evolved from point solutions to applications, ultimately leading to system-level change. For example, when electricity simply replaced steam within the same system (point solution) the benefits were limited. When electricity drove the system-level redesign of factories, the game changed.
There is a lot to learn about the progression of AI by observing the path of previous general-purpose technologies (GPT). Introducing AI into an existing system will add value — but the greatest value is only attained when the GPT forces a change to the system. Had electricity not evolved from replacing steam (point solution) to providing individual motors to machines (application), Henry Ford’s reimagining of production would never have occurred. The evolution of electricity through this cycle took forty years.
By 1880, it was clear that electricity had great potential to improve how factories operate. But it required another forty years to understand how to design a factory system that took advantage of electric power. To the best of our knowledge, no one envisaged the ultimate systems that were built of electricity. The process of discovery took time, as people evolved in their understanding of what electricity could do. With AI, we are closer to 1880 than 1920.
Ajay Agrawal, Joshua Gans, Avi Goldfarb — Power and Prediction
As AI moves towards general-purpose technology status, the comparison to electricity is instructive. This recent article touched on similar system-level themes as it relates to healthcare. The article explores how artificial intelligence will change the regulation and organization of medicine. It looks at the alternative paths ahead and the approaches to the coming organizational challenges that AI presents for modern medicine. Three possible scenarios come into focus: the first is a priority to ensure that tools serve the physician, and the physician remains in control. The second is to recognize that a radical transformation in health care is underway and to adapt accordingly. The third comes via a direct quote from the article:
A third strategy, perhaps a middle ground, is to think carefully about what health care providers can do that machines cannot. Many of the informational, operational, and maintenance services that physicians provide might be replaced by machines, but machines will never be human. Therein lies some kernel of irreplaceability that human providers (not just physicians) can and must offer. Ironically, physicians played this role historically. Before medical science armed them with a wealth of new tools, physicians tended to the dying, comforted the sick, and served as a compassionate custodian of health. But the profession moved away from this role as the medical science exploded with new capabilities.
Anthony Weiss, Luke Sato, Barak D. Richman — How AI Will Change the Regulation and Organization of Medicine
But if we shift back to the book, this third strategy of moving back to being compassionate custodians of health depends on system-level change. The authors state that to achieve this worthy goal, we need more than new AI technologies. We need a new system, including new incentives, training, methodologies, and culture for doctors to utilize their technological tools in the manner aligned with this aspiration. The complexity of the health system works both against AI and system-level change. Per the book, at the end of 2019, health care had a smaller fraction of jobs involving AI than every other industry except construction and arts and entertainment. Even accommodation and food services, and transportation and warehousing, involved more workers with AI-related skills.
If we view this discussion through the lens of possibility (both disruptors and opportunities) and in the context of the evolutionary path of prior general-purpose technologies, then the possibility space spans point solutions, applications, and system-level change. This progression moves from low to high as measured by disruptive potential and value. Said another way, disruptive potential is at its highest if a general-purpose technology like AI leads to system-level change, as is value attainment.
AI applications in medicine include disease diagnosis, automated surgery, at-home patient monitoring, personalized treatments, and drug discovery and repurposing. These opportunities have created worries about a “dark side to AI in health care” where AIs compete with MDs for diagnosis.
Ajay Agrawal, Joshua Gans, Avi Goldfarb — Power and Prediction
Given the system-changing potential of AI, the future of health and medicine could be very bright. However, it took forty years for electricity to change the factory system, will it take another forty to change health and medicine? Will the complexity of this system undermine the possibilities? As leaders in health and medicine explore the impact of AI, it would be wise to consider the evolutionary path of previous general-purpose technologies. The authors believe that AI could have a bigger effect than previous generations of general-purpose technologies, from the steam engine to the Internet. A large portion of the audience I polled believed the same. When asked the following question, 48% of respondents answered in the affirmative: Will artificial intelligence be more profound than fire, electricity, and the internet?
As leaders assess the implications of artificial intelligence, it is important to acknowledge both the magnitude of potential impact, and the time horizons driven by the need for system-level change. To add to this complexity, we now need to consider the implications of a fundamental change to society’s organizing system. Our evolution towards ecosystems introduces a system of systems dimension to this historical view of general-purpose technologies — for a different day. I highly recommend the read and added the book to my library.
Originally published at http://frankdiana.net on June 27, 2023.
