An Update on IBM Watson for Health Care — Ignore The AI Carnival Barkers
Reading time: 2 minutes 30 seconds
Check the news on IBM Watson for Health Care:
- STAT reports that IBM is halting sales of Watson AI for drug discovery
- FierceBiotch reports that IBM employees were recently told of this decision and are reassessing Watson’s prospects in the broader biopharma industry.
- Troubles in IBM’s Watson for Oncology – a clinical advisor project – became visible in 2017.
The root of these negative news events is not a failure of technology. It’s a failure to meet the inflated expectations that people had for future versions of the technology.
Watson – including Watson AI for drug discovery and other Watson Health initiatives – delivered some notably positive results reported in peer-reviewed academic research and earned scores of citations by other researchers.
But Watson missed on the
- Magnitude of the effects people expected Watson to deliver
- Ability of the market to provide the required training data to feed to Watson
- Readiness of external upstream and downstream systems to integrate with Watson
- Level of comparative advantage produced by the effects on the processes in question
- Resistance to adopt the resultant solutions
Many projects IBM and its “joint venture” partners took on contained elements that no one had tried before. They took it on faith they’d find all the problems and fix them all. They failed to differentiate between publicity and delivery. They took big hits for failing to handle the known and unknown unknowns.
In dropping the Watson AI for drug discovery program, IBM made a business decision that the drug discovery program’s return to IBM was going to be less than the return they would get from other IBM Watson projects. Net: it became a commercial failure. Drug discovery wasn’t producing enough commercially.
I’ve seen announcement after announcement of “joint ventures” to do what heretofore couldn’t be done, built on a framework of marketing claims rather than good hard proven science, engineering and operational results that already demonstrated it could do what was hoped for.
It was a confusion of ambition over achievements, supercharged with super-strong marketing and sales as well as a muscular PR strategy that took advantage of well-developed relationships between firms.
The Watson for Drug Discovery and Watson for Oncology disappointments are reminders that big-bang projects can fail big time and first-mover advantages can turn into first-failure disadvantages.
I am absolutely convinced that:
- There will be hundreds of thousands of applications of AI technology that can make dollars and cents and provide competitive advantages in terms of improved efficiency, greater insight, better customer experiences and deep competitive advantages.
- Businesses need to invest (and take on risks) to achieve substantial strategic objectives.
I’ve found very few non-tech-industry CEOs who, in the stress of a tough analyst earnings call, will declare that their top strategic objective is to become an AI company or a digitalized company. Sure, you’ll hear them scatter those words but that’s because they’re in the zeitgeist – in the press and on the lips of most consultants.
Ignore hollow hype, as in marketing claims to “Embrace the Cognitive Era” and nonsense, as in “if you don’t invest, how do you know your competitors won’t beat you to market and bury you because of that.”
Assume a position exactly opposite of the carnival barkers and circus ringmasters urging everyone to run over the cliff into wanton technology spend.
Become the antithesis of the people yelling go-go-go and buy-buy-buy AI (and digitalization.) Counterbalance the fever! Stay rational.
This post reflects my current opinion. No one paid for this post.