#DataScienceFail?
Only 12% of enterprises have deployed AI at scale. That’s DOWN from 15% in 2020 according to the seminal survey on data science in the enterprise by Tom Davenport and Randy Bean at New Vantage Partners. Why?
In the same report, 90% blame culture, not technology.
In other words, more data science isn’t the key to doing more data science. Culture is the key to getting more value from AI.
You might be saying to yourself: Aha! Culture! Why didn’t I think of that! Gotta get me some of that culture!
Or, if you think a lot about this stuff, you’re thinking: “Sure, but culture is hard.”
* * *
Three things can help: education via case studies of cultural excellence in AI, culture design methodology, and tools that facilitate data science teams.
In 2021, you’ll find a stream of case studies about data science excellence on this blog. For example, 50,000 Data Chiefs is about Panera Bread’s culture of data curation. It starts with CIO John Meister. While many CIOs hoard data, Meister democratizes it. He put responsibility for data in the hands of the business.
Jeff McMillan at Morgan Stanley democratizes his data science team. He tethers data scientists to business people. His enlightened AI culture is among the best I’ve seen.
Innovative cultural design methodology is exemplified by Gapingvoid. Their art, which is part of their culture design program, inspired this post. Gapingvoid uses social science, behavioral science, neuroscience, management science, art, and tenacity to affect change.
In the 1900’s, public libraries helped democratize knowledge by making it freely accessible to everyone, not just a few privileged elites. Model Ops is like a public library for data science. It’s a set of tools that democratize access to AI, which in turn makes cultural change possible.
2021 will be the year of Model Ops. Not because it will help data scientists make more data science; because it will help create better cultures.
