The goal of this piece is to continue on the topic of implementing AI and continuing the digital transformation in your business and organizations.
We’ve talked about identifying specific problems to solve and identifying different initial team roles. Next, let’s focus again on Data.
The Initial Team Is Built – What’s Next? It’s All About the DATA
In this Substack, we’ve talked about the first team that will handle your initial AI Journey.
What goes right along with that? It’s your Data Preparation.
The famous Stanford professor and AI entrepreneur Andrew Ng calls data “the new electricity”. And he’s right.
No machine learning (ML) or AI project really gets off the ground with bad data.
What’s that mean?
Well, it could be bad labeling or annotation. It could be data sets that are in disparate places in your business and stored in different systems or managed by different people with their own nomenclature.
Or, you might just not have enough of it.
Currently, ML models operate best with many iterations using lots of data. There are companies working on software that one day won’t require vast datasets (or coding ability), but generally, the more data the better.
How Important Is It?
In the book Implementing AI Systems by Tom Taulli, he includes a great example of how valuable data is.
“In late 2015, IBM announced the acquisition of the Weather Company—which included the weather data, forecast information, website, app, and various intellectual property. The price tag was over an estimated $2 billion. What was the rationale for this? Was IBM getting into the weather business? The deal was actually great for the company. Data is the fuel for AI, and weather data has broad applications. IBM folded the Weather Company assets into the Watson AI platform. The forecasting segment, called WSI, was likely the most important asset. The business included license revenue from over 5,000 companies in industries like airlines and utilities. In terms of the data assets, they included three billion weather forecast reference points. There was also infrastructure for data collection from over 40 million smartphones and 50,000 airplane flights per day.”
$2 Billion big ones – for data that the Weather Company had been collecting as part of day-to-day operations.
And IBM wasn’t the only suitor for this vast trove of data points.
Many others saw the value of what the Weather Company had built. As Tom says, data is the fuel for AI.
How Did We Get Here – Hasn’t Data Always Been Important?
Yes, it has. But with the explosion of computing power, the transition of all documents and data online, and the Internet of Things (IoT), there will be more data generated in the next 5-10 years than the previous 30, according to research from IDC.
Think of IoT as your devices you use in daily life -- smartphone, GPS systems in your sweet new ride, your security cameras on your McMansion in the burbs -- all of these devices have sensors and software that are capturing information and data points.
Now multiply that exponentially by hundreds of millions of users and you can get an idea of the volume we’re talking about here.
This is great for AI.
So - value the data in your organization. Centralize it, structure it as best you can. Then, plan AI projects around it. The future awaits.