Tom Merritt breaks down a recent Forrester report about data science, machine learning and AI. Here are the five things you should know.
Whether it’s machine learning, analytics or data mining, tech pros need to make strategic decisions about data science. However, few of the decision-makers have direct experience with the relatively new and definitely complex discipline. Forrester recently put out a report called “The Tech Executive’s Primer On Data Science, Machine Learning, And AI.” While you should read the whole report to get fully up to speed, this can help you get started. Here are five things to know about data science.
SEE: Navigating data privacy (free PDF) (TechRepublic)
- If it looks like it does in the movies—though impressive—it’s probably not AI. Actual AI products that can actually do what they say they’ll do, have limited intelligence and autonomy. It’s like the old adage, “If it looks too good to be true, it probably is.”
- Worry first about humans. Any algorithm is only as good as the data it’s given and the data is given to it by humans. Social issues like race and gender get the most attention, but there are many other kinds of biases that can affect data sets as well, and none of them lead to a good outcome for your business. Screen your data, and test and validate your models.
- Improve data over time. Just because you’re trying to make the dataset better, doesn’t mean you should hold off starting the project. Working with data is what lets you learn what data you need and in what form. That also goes for your algorithms.
- Choose projects you can actually implement and measure. Having an end result in mind is not enough. Using an algorithm because it can do a thing, doesn’t mean it’s the right thing. Know what you want to achieve, that you have the right system to achieve it and how to measure your progress.
- Think of the users. You can have the best data, best algorithm and best system in the world, but if the users don’t know or don’t care about using it, it won’t matter. Involve end users right from the start.
Machines are still learning, so we need to be smart about using them. Hopefully these tips get your head pointed in a good direction. For more, check out Allen Bernard’s article: 7 best practices for implementing data-driven technologies, like AI and machine learning.