top of page

Designing an "AI First" Company Means Starting with AI and Then Plugging in Humans

I wrote my book The Way of the Navigator in 2013 to help business leaders and policy makers understand many of the ways AI would impact business, the workforce and society in general. I spoke quite a bit about how "this time actually was going to be different" than past technological disruptions because the word “intelligence” in “Artificial Intelligence” actually meant something. AI will impact and disrupt and change "everything" in so many ways that I won't go into here, but I do want to talk about something critical that I tried to highlight in my book and that is getting lost in the discussions around AI.


Contrary to most commentary on the subject of AI, it won't just be low level jobs that face automation, but in fact, AI will take over many of the jobs that require the most education, training and expertise. I argued then, and continue to argue now, that the failure to truly understand this and then design around it means that any digital transformation effort that you attempt as a business leader will fall far short of realizing the full opportunity that AI offers.


Once we understand this we logically and quickly come the realization that in order to "digitally transform" a business, or start a new one, we need to first understand what AI can do at every single step in the process and then build human support around that. In other words, design a system that I call "AI First" and then figure out how, where and when to plug in humans. Most Digital Transformation efforts fail because they do not start with the technology as design.


Let's say you are the CEO of a large healthcare company and you decide to pursue digital transformation. Like many other CEO's you will direct your staff to digitally transform your company, and like most companies when it is all said and done you will have tinkered with AI on the edges ("Look, we have chatbots!") and sent many of your employees to Design Thinking seminars and automated some of the lower level tasks. When you should have been asking your team "How does AI allow us to completely re-think our organization and how we deliver services and care?" And by doing this you might even discover that your core business model changes as certain profit centers disappear and other areas open up.


How would you potentially re-design your "healthcare system" to provide better service at a dramatically lower cost? I am not a healthcare expert but here are some off the cuff ideas. I would design a system that leverages wearables for constant monitoring of my client, create robust knowledge bases to tap for diagnosis and for predictive analytics, DNA testing, monitoring of unstructured data tagged to geography (to determine flu outbreaks, drug overdose stats, etc.), intelligent virtual assistants for patient inquiries and to explain test results and much more.


And once this system was designed I would ask "At what point should my patient talk to and/or meet with a person and what is the skill set of that person?" As an example, let's look at radiologists. Radiologists are some of the best trained and highest paid medical practitioners. But in many recent studies AI detects disease more accurately and at a much faster pace. Or when should a patient actually see a doctor? Most of the time when I visit my doctor we both spend our time googling things that I think I have to see what the symptoms are and what the treatment is. Can a nurse on the phone with access to all of the robust technology I described above do most of the work a doctor does today? Let's say they can do 50%. And can virtual assistants answer 20% of a patients questions? That means we need less than half the doctors we do today or we can serve more than twice as many patients. Or what radiologists? Radiologists are some of the best trained and highest paid of medical practitioners and yet in many recent studies AI detects disease more accurately and at a much faster pace. So can a radiologist working with AI soon do the job of 10 radiologists? 100? And the AI will only get more "intelligent" not less. For example nurses co-working with AI will soon take over 90% of what a GP does does, and yet, at the same time virtual assistants will continue to improve and will take over an ever increasing amount of tasks that the nurse does. And this process will continue until their is a "cognitive" or "intelligent" system in place with much fewer "humans" and each of these humans doing jobs probably much different than today.


The reason it is important to start with an "AI First" design approach is because if you build a system that anticipates this shift from human to machine (100% doctor today to 50% doctor to 50% nurse ratio tomorrow and then to 20% doctor to 50% nurse to 30% virtual assistant ratio the year after that, etc) you can set expectations with your employees and retrain them for the future. You can be proactive instead of reactive.


By starting with an "AI First" approach we can quickly see that we need many fewer of the highest paid and most trained people on our staff. Technology replaces them. The reason it is


One problem of course is that the highest paid and most trained people on your staff are also the ones who typically make the department level decisions around "digital transformation" and so therefore without oversight you will never be able to truly digitally transform your organization because they will never let themselves be automated out of a job without a clear path to something new.


Transformation through AI and building from the ground up is going to be a challenge for large companies. It might actually be impossible. But if it is not attempted then a start-up in Silicon Valley designing a company from scratch that thinks how things can be done with AI from the beginning.


Comments


bottom of page