Artificial intelligence (AI) has the power to surprise, delight and frighten. It seems almost magical. But this false appearance can tempt organizations to skip steps in their data strategy. Here’s how it happens, and how to solve it.
Step 2: AI-Powered Enterprise.
Step 1: Connect printer to computer
I often think about how many organizations can’t do the data equivalent of hooking up a printer to a computer. There are many steps to get down before AI, as the Data Science Hierarchy of Needs illustrates:
Logging, collecting and moving data comes first. Without solid rudimentary data pipelines, an AI initiative is dead in the water. As an HBR article puts it:
… companies that rush into sophisticated artificial intelligence before reaching a critical mass of automated processes and structured analytics can end up paralyzed.
Nick Harrison and Deborah O’Neill, “If Your Company Isn’t Good at Analytics, It’s Not Ready for AI” (source: Harvard Business Review)
Most firms have admitted they’re not ready
So, maybe organizations just think they are ready for AI, but they aren’t. I’m not even sure this is true. Take some staggering numbers from the data consultancy NewVantage Partners:
- 71.7% of firms report they have yet to forge a data culture
- 69.0% of firms report they have not created a data-driven culture
- 53.1% of firms state they are not yet treating data as a business asset
- 52.4% of firms claim they are not competing on data and analytics
If organizations admit they’re not competing on data now, what makes them think jumping the gun to AI will fix things? I have an idea.
It sounds like magic, so it must be easy
I think it’s because the more advanced a technology is, the more magical it becomes. And we tend to think magical-looking things are effortless, because that is how they appear to us.
Most manager can imagine the tedium of compiling spreadsheet-based reports, and building a reliable data flow. They probably know that it’s anything but magical to get a data warehouse up and running.
AI, on the other hand? Not even the authors can audit how those algorithms work. It works with spellbinding accuracy, and makes far fewer mistakes than that hatchet-job weekly spreadsheet report. So it must be easier, right?
Data upskilling takes work
Not so fast. That hatchet-job spreadsheet indicates not that you should accelerate to AI, but that you should clean up the processes and definitions that are blocking the automation of that report. Then automate the report. Then do all the other stuff on that Hierarchy of Needs. And then do AI.
Don’t think that AI is magic, or it will sink your data efforts.
So how do you get from an organization that can’t do the data equivalent of connecting their printer? I think it’s a data academy.
Leave a Reply