The big takeaway from Epoch AI’s blog article "Most AI value will come from broad automation, not from R&D" is that the transformative gains from AI won’t come from a handful of research breakthroughs as leaders such as Sam Altman, Dario Amodei, and Demis Hassabis are claiming—it will come from systematically automating the everyday work that runs the global economy. R&D has driven only ~0.4 percentage points of U.S. total-factor productivity growth since 1988; broad labor, by contrast, accounts for roughly five times the output elasticity. Once a model is smart enough to replace a lab scientist, it’s already more than capable of streamlining sales to delivery processes for a design engineering department, managing the financial services client lifecycle, or triaging revenue-cycle claims—tasks that soak up orders of magnitude more spend and generate orders of magnitude more value.
This is a passion of mine and I’ve tried to make the same argument in practice for UKG Pro and Process Street:
-
In “Improve Process Accuracy and ROI Using Process Street” I showed how checklist-driven automation eliminated billing rework and unlocked cash for a healthcare RCM team.
-
“Process Optimization with UKG Pro and Process Street” walked through integrating workflow engines into HCM so frontline managers stop fighting spreadsheets and start directing talent.
-
In “Hospitals and AI. Tools for Success” I warned that rushing shiny AI pilots into production without hardened processes invites hallucinations and compliance risk.
What this means for leaders: Stop treating AI like an R&D moon-shot and start treating it like lean manufacturing for knowledge work. Map the workflows that drive margin; instrument them with process intelligence; and then let domain-specific automation do the heavy lifting while your experts tackle edge cases and innovation.
That’s exactly the lane our Process Intelligence team lives in—capturing hidden process debt, surfacing data you can trust, and orchestrating automation so your people can focus on higher-order problems. As the economy pivots from “build the model” to “scale the workflow,” the organizations that win will be those who understand their processes better than any model possibly can—and have the discipline to automate them end-to-end.
Comments