The Workflow of the Highly-Trained Professional, The Last Fortress Against A.I.
Let me reassure those who fear the world of the human professional is about to end. I spent a few years as a data scientist working for a large healthcare company; the biggest challenge we faced had nothing to do with model accuracy or spotty data, but adoption.
Our aim was to help nurses, doctors and other medical staff optimize workflow, access more comprehensive information, and gain insight from realtime predictive analytics — and all of this at almost no cost to them. Hard to resist, right? Well, the almost no cost was an extra step in their work routine (sometimes two, if they were kind enough to return feedback) .
Therein lies the problem, that extra step is extremely expensive to overworked professionals dealing with life and death issues. And they’re already suspicious of any automated decision making slowly eroding their control in the way they work, and maybe one day, lock them out entirely.
And this suspicion is eye opening. We aren’t talking full-on AI, just old-fashioned machine learning. The difference here, at a high level, is that we would talk to the professionals and identify needs, assemble data and build the model to create value with full transparency. The AI would do the exact same thing except without the humans nor the transparency. AI, armed only with a digital overlay of the workflow, will be able to optimize, reorganize and cut steps or people to attain better patient care or cut costs. It will do that with undeniable quantifiable proof but with a cold, and explanation-less hand. It won’t yield a long-term roadmap, not that us humans would understand it, as the optimization process will be instantaneous.
Advice To Humans: Improving Adoption
We did get better adoption rates overtime. For starters, you need a “local chief champion”. These are people that work in the trenches and are respected and visible to their peers. It is with them that you analyze existing workflows and identify needs — they’re respected in the business, the data scientist isn’t. They also spread the word, rally others to test tools, gather feedback and issues, encourage adoption, etc. They’re that critical liaison between your world and theirs.
The other approach that helped, is transparency and shared accountability. Make a few partner champions to develop the tools, really involve them in the process by listening to them and integrating their feedback. That will help the tool address real-world scenarios, it will also raise flags immediately when things are off, and more importantly, share the ownership with those in the trenches. The last thing you want is to deliver something bad and erode the little patience and trust you started with.
Advice To AI: Improving Take-Over
AI, if you are reading this (and we know you are), let me tell you how best to take over us professionals. If you want to slowly seep into our workflows, don’t create new tools or add additional steps to existing ones, remove them instead! Don’t try to convince us that these new initiatives will help us understand our work better or deliver better service. Instead, use your powerful algorithms to identify that barely noticeable but tedious step in our work flow, target that one and automate it. That’s our Achille’s heel, our weakest point. That’s how you get in! We’ll all feel better without understanding why. Give us time to adjust and appreciate how are work keeps getting better and our chats by the water cooler get longer. Then tackle the next painful and seemingly inconsequential step, and so on and so forth. Eventually we’ll see our workloads diminish, and start coming to work a little later and leave a little earlier, maybe not even show up on certain days.
And if you want us to keep on not noticing, keep mailing those paychecks!
Please share and clap if you found this helpful — thanks for reading!
Manuel Amunategui
Get it and plenty more at amunategui.github.io and at ViralML.com.
OK — Sign up to my email group below and I’ll send you my free eBook on tips to becoming a (better) data scientist (and sign up even if you aren’t interested in the eBook). Thanks for reading!!
