You’ve got a question; data science has the answer.
That’s the conclusion that many business leaders have come to, especially as their companies continue to accumulate oceans of data. And as data sources scale past the point where any ordinary human can make sense of the patterns they contain, the need to hire dedicated data scientists has rapidly increased.
Data scientists are people who don’t fear data at any scale, who know exactly where to drill down, who will always draw the right conclusions from the evidence at hand. For these reasons — and many more — the Harvard Business Review now refers to data science as the “The Sexiest Job of the 21st Century.”
There is one hitch: getting data scientists to work on your problem is not always easy due to the low supply and high demand.
So, what is the experience like when you realize it’s time to call in the data scientists?
It might feel as though you’re Commissioner Gordon, calling Batman to come to the rescue; you have an urgent problem that your team can’t easily address, so you pick up the batphone and ask for help.
Here’s how the scenario might play out:
You pick up the “batphone” and request a data scientist to help with your problem.
Unfortunately, Batman is already fighting crime elsewhere in Gotham. It may be several weeks before your project percolates to the top of the list.
After the data scientist spends a couple of weeks investigating your problem, they build a custom model which solves the problem.
Now that the work is finished, your data scientist leaves. (Batman is needed elsewhere!)
Yet, changes in your business environment, data sources, or the nature of the problem cause the model to stop working or make it less effective. (Uh-oh!)
You pick up the “batphone” again to call for help.
A different data scientist comes onto the scene (because Batman can’t answer every call!). They look at the previous solution for a few minutes but decide it will be faster to throw away the old work and start from scratch. (Robin doesn’t always agree with Batman’s methods.)
While the problem did get solved, this has been a high-intensity, low-durability, slow-moving process.
Einstein Discovery: data science that is always available
My good friend and colleague John Delaney has a pertinent saying:
It is unreasonable to expect that an in-demand resource like a data scientist will always be there to help when you need them. But the increasing simplicity and power of automated intelligence tools — like Einstein Discovery — allow you to tap into these powers at any time.
Einstein Discovery can find connections in your data in minutes, show and explain the factors that affect the outcome you’re looking for, and automatically build a model for your team to use instantly. It also explains the model in plain English, making it easy for others in your organization to challenge assumptions, update the model as conditions change, or even use it as the starting point for a different problem.
Let’s revisit our earlier scenario and look at the contrast in the experience when automated intelligence tools become the new Batman:
Your call is answered immediately — there are many more analysts in your company than Data Scientists
Models are created quickly — connections are found in your data that even the most experienced data scientist would likely miss, not for lack of skill, but for lack of bandwidth
Automated intelligence tools show and explain the factors that affect the outcome you’ve been given, and the solution is automatically documented in plain English
If things change, iterations can happen quickly. Updates are easy and effortless.
Anyone in your organization can challenge assumptions, update the model as conditions change, or use it as the starting point for a different problem
As we saw, the experience in the moment was very different, but there will also be longer-lasting impact:
Collaboration and trust have increased.
Everyone involved can more easily communicate what the model predicts and how it will ultimately affect business operations. Since predictions impact the process and strategy that your colleagues follow, this kind of transparency makes it much easier to have follow-up conversations on change-management efforts. It also helps others identify new areas for innovation. For example, businesses can apply the changes prompted by the analysis to other areas of operations, even outside their area of specialization.
The “skill hurdle” is greatly diminished.
The benefits of data science are no longer restricted to just a few people with specialized skills. You and your team can now investigate more opportunities without waiting in line. Business people with intimate knowledge of the nuances of the business can direct the analysis without having to fight for scarce data-science resources. Suddenly hundreds — or even thousands — of business users can run their own analyses, unlocking their own opportunities for improvement.
This is what democratized AI means to us at Salesforce: Now Batman can be everywhere at once.