AI and machine learning are taking the world by storm, and that includes the world of marketing. AI-powered software platforms for marketers are also growing in popularity, but a report by Forrester (article is behind a paywall) suggests looking—and learning—before you leap.
Everyone has a notion of what artificial intelligence is. We use Alexa, Siri, and Google Assistant every day, read about self-driving cars, and hear about how automation is replacing everyone with thinking robots and we can't help but form a concept about what AI is—and it's one that is often incorrect.
Knowing what AI is, Forrester's report says, is a key part of making the right decision to truly make it work for your business.
SEE: Modern Marketing & Entrepreneurship Bundle (TechRepublic Academy)
The report may focus on how B2C marketers can make use of AI but anyone considering an AI platform for data analytics can learn from the five AI myths Forrester attempts to dispel.
Myth 1: AI is new technology for marketers
Not true, says Forrester. In reality, demand-side platforms (DSPs) have been applying machine learning techniques to programmatic real-time bidding (RTB) for years. They're doing it on their end, though, so all you see are optimized ad campaigns.
Outside of marketing, those interested in using machine learning to accomplish a task should always look around to see who's offering it—just because someone says they're the only one on the market doesn't mean they're actually offering something unique.
Myth 2: AI is all about the algorithm
Nope—it's the data. AI's current strength is in its ability to analyze data and find connections that humans miss. The best algorithm in the world can't do much with data that's sparse, poorly organised, or inaccurate.
Don't trust anyone who says their algorithm can do more with your data—if you give them the same stuff you give someone else you'll get the same results.
Myth 3: AI platforms work out of the box
That shiny new learning machine you just set up isn't ready for the big time: It's a new mind with the framework for complex problem solving but none of the experience it needs to draw conclusions.
SEE: Understanding the differences between AI, machine learning, and deep learning (TechRepublic)
AIs try to emulate human cognition, and in order to do so they have to be trained. Forrester reports that it can take weeks to organize data, days to create training models, and up to six months to fully optimize algorithms.
In other words don't expect immediate results.
Myth 4: AI autonomy will kill jobs that rely on it
Not so, at least according to the report. What AI does (both for marketers and other data-driven professions) is free up time spent on minutiae. Let machine learning do all the number crunching, data compiling, and report generating so your human employees can close sales and create content.
Myth 5: AI findings will be rich with customer insights
Learning machines don't care about what they learn and they don't understand a bit of it. That means that while an AI may discover a trend between two previously unconnected data points it has no idea how to break it down into a real, intelligent insight.
Drawing insights from data is left up to us humans, and in many cases we can only hypothesize on the connection. Asking the AI is useless—it's not smart enough to know what the data it processes represents, and it's not human enough to care about it either.
- Machine learning: The smart person's guide (TechRepublic)
- How to Implement AI and Machine Learning (ZDNet)
- Research: Companies lack skills to implement and support AI and machine learning (Tech Pro Research)
- Machine learning adds punch to predictive analytics (ZDNet)
- 5 big data trends that will shape AI in 2017 (TechRepublic)
Brandon Vigliarolo has nothing to disclose. He does not hold investments in the technology companies he covers.
Brandon writes about apps and software for TechRepublic. He's an award-winning feature writer who previously worked as an IT professional and served as an MP in the US Army.