According to a new report from big data management company Talend, 2017 will not be the year of AI, but will actually be the year of (wait for it) real-time analytics.
Yes, real-time analytics. (After all, it has to be the year of something.)
This finding, which is based on a survey of 189 people who attended a Talend event in Paris last fall, may run counter to the prevailing wisdom that AI and ML are taking off in big ways, and that they are doing so right now.
Actually, it supports that theory, in a way. Bare with me.
Part of the problem is that too many people are throwing around the term AI, and it’s getting watered down. Strong AI, or True AI, is defined by Wikipedia as “a machine that exhibits behavior at least as skillful and flexible as humans do.” True AI is the type of stuff you see in the movies, according to a recent Atlantic article, and not rudimentary filtering from yet another software program (YASP!).
Real AI–like cars that can drive themselves, drones that kill terrorists by themselves, healthcare programs that can diagnose and treat diseases, or robots that can make us think they’re real–will indeed disrupt human life on a massive scale, but only when it actually becomes available. We’re still a ways off from that world (or WestWorld, as it were).
In the meantime, CIOs are placing their bets on big data tech available in the here and now.
Here’s how the big data chips are lining up in the minds of CIOs, according to the Talend survey, in order of importance:
- Real-Time Analytics: 26 declared it top priority;
- Master data management: 20 declared it top priority;
- Self-service data preparation: 18 declared it top priority;
- Artificial intelligence/machine learning: 10 declared it top priority;
- Internet of Things (IoT): 5 declared it top priority.
While AI/Machine learning and IoT are on the radar of many CIOs, they rank much lower in importance, says Ashley Stirrup, chief marketing officer for Talend.
“While dazzling proof of concepts will continue to capture companies’ imaginations, IT leaders will quickly realize that AI/machine learning are a lot harder than they appear at first blush and a more measured, long-term approach to both is needed,” Talend tells Datanami.
“CIOs today are not only challenged to drive their organization’s digital transformation, but also spend time and budget maintaining day-to-day business operations,” he continues. “As a result, mainstream business functions such as real-time analytics and self-service data preparation—functions that more immediately make employees more effective in their roles—need to take precedence, at least for the near term.”
This finding jibes with what the respected injury analyst group Gartner has been saying too. AI technologies like machine learning, cognitive expert advisors, smart robots, and autonomous vehicles were all at the top of its latest Hype Cycle last August.
That means they’re at the height of the “Peak of Inflated Expectations,” where the hype is so thick, you can cut it with a knife. “Smart machine technologies will be the most disruptive class of technologies over the next 10 years…” Gartner says.
Technology is moving so quickly right now, and few have the confidence to go out on a limb and predict exactly what we’ll have 10 years from now. Over the past 20 years, Microsoft founder Bill Gates, ever an optimist, has been consistently wrong about with his predictions on imminent breakthroughs in the capabilities of natural language processing (NLP) and speech recognition, which along with computer vision (object reception and facial recognition) are some of the core AI problems that need to be solved before we can get real, or strong, AI.
The good news is the really smart guys at Google, Microsoft, and IBM—as well as the top research universities around the world–are funneling billions of dollars and man (and woman) hours into solving these problems. We’ve made a lot of progress over the last few years with advances in deep learning, and particularly, in building very large neural networks that can train on huge data sets, and get us closer to a reasonable facsimile of how human brains process knowledge and learn.
But, we’re not there yet. Less than a year ago, Facebook CEO Mark Zuckerberg predicted that AI would indeed surpass human understanding, but not for at least another five to 10 years. Computer vision and NLP just aren’t there yet, he said.
“…[T]oday our systems can’t actually understand what the content means,” Zuck said in a conference call. “We don’t actually look at the photo and deeply understand what’s in it or look at the videos and understand what’s in it or read the links that people share and understand what’s in them. But in the future we’ll be able to, I think in a five or 10-year period.”
In the meantime, real-time analytics provides a possible payback on a much shorter timeframe than true AI. You don’t need advanced machine learning skills to be able to make decisions on incoming bits of data—in fact, simple SQL will suffice. With emerging open source tools like Spark, Flink, Beam, and Kafka, you can create advanced data pipelines that allow you to process data as it arrives.
And you can even use some machine learning techniques to process the data more intelligently in real time. It may not be true AI. It may be a slightly more intelligent filter combined with a dose of SQL processing and heaps of rudimentary programmatic logic. It might be YASP. But if it helps your business achieve its objective, who cares?