IBM Watson's Health Saga

Dated Jun 25, 2018; last modified on Thu, 02 Sep 2021

  • IBM acquired Phytel, Explorys and Truven for their troves of healthcare data and proprietary analytics.

  • From 2010 to 2015, IBM has spent +$15bn and Watson is still not profitable.

  • Marketed abilities: scan literature & patient history to recommend cancer treatment plans, offer preventative medicine & medical imaging.

    • MD Anderson spent $60m on a Watson project before concluding ‘not ready for human investigational/clinical use’.
    • Still, nobody ever gets fired for choosing IBM.
  • Watson is an amazing information retrieval system, but marketing insists on overselling it as AI

  • IBM’s product department pivots too much and has non-technical leadership. Watson isn’t even in the product dept’s planned services.

  • IBM is losing talent. It won’t win AI. But the CEO still talks of Watson’s Law and exponential benefits.

  • Getting data isn’t hard, but most of it is in prose form. Good luck cleaning and structuring it. It’s hard to find domain experts who also know NLP to guide the process. Garbage in…

  • Finding the next fool: CRUD app running on Azure with a D3 reporting dashboard == “revolutionary AI-driven real estate analysis product”.