With the enormous wave of machine intelligence technologies comes yet another hype curve of products that cover virtually every key product segment in the industry.
Machine intelligence (MI) is no longer the sibling of artificial intelligence or what was pejoratively referred to as “machine intelligence as a magic box.”
When MI is paired with Internet of Things (IoT) and big data deployments, there are precious few things in enterprise or government computing that aren’t touched by these platforms. In this environment, it becomes more difficult to say what MI is not rather than what it is. (You’ll see what I mean if you click on this MI Landscape chart from Shivon Zilis.)
Exacerbating this fuzziness is the challenge that machine intelligence has had such a checkered history, leading many to wonder whether this iteration, unlike its predecessors, is going to stick.
Historically, the gripe was that machine intelligence had no real mainstream application that would lead to consumerization. If you asked most people what machine intelligence was, they’d likely refer to Big Blue beating Gary Kasparov at chess. Just what every enterprise or government needs!
But now, many VC’s are betting that “MI 3.0” has legs, largely because of the underlying technologies that can now feed into it. Vast and disparate IoT feeds sent to central (big) data repositories create an enormous source of fuel for machine intelligence that can instantaneously convert signals into insight.
David Moschella, Research Fellow at CSC’s Leading Edge Forum, says that MI is no longer driven from the top-down as with the old “expert systems” movement.
“Just as we saw with PCs, networks, online services and smartphones, consumerization is now transforming MI into a bottom-up dynamic. This time, the driving force is not technology; it’s Big Data. Many of the most important machine intelligence initiatives today – such as language translation and image, facial, activity and emotion recognition – are based on predictive analytics that get more accurate as the data sets get richer, and consumer markets are where the biggest and best data resides.”
As proof, look at augmented and virtual reality tools that have a rapidly increasing following in business, government and consumer spaces, far beyond their previous “eye candy” or “magic box” status. Pop culture has accelerated the engagement, as seen in the 20 million daily Pokémon Go users.
The consumerized MI movement is getting greater traction as a result of the burgeoning quantified self movement, or in layman’s terms the Fitbit Movement. Five years ago, we would have never regarded a personal exercise tracker as “machine intelligence,” but companies can use feed the anonymized data from individual IoTs into a big data repository to derive insight.
So what can enterprise and government IT leadership do to increase their machine intelligence quotient? Here are four first steps in that direction.
- Conduct MI Opportunity & Normalization Audits
The return on investment for machine intelligence deployments can only be determined after aggregating your data feeds. You might determine that there simply aren’t enough data sources to produce meaningful, actionable insight or predictive analysis at this time. With many enterprise applications, such as Salesforce or Marketo, it is possible to “grow into” machine intelligence. But far too many firms have invested in and underutilized MI platforms thinking they had more data than what’s actually there to start.
- Determine what insight you’re looking for before the deployment
This may sound like a blinding glimpse of the obvious. Data scientists can surely unearth deeper insights that might not be visible to the naked eye, but to start, you should have a core set of “dream data deliverables” that will add competitive value to the enterprise. Far too many enterprises and governments start the process with “I don’t know what I’m looking for so can you tell me what I should know?”
- Do an internal skills audit
Many public and private enterprises enter into machine intelligence projects totally underestimating the skills necessary to get meaningful insights for competitive advantage or cost savings. A successful project requires ongoing investment in employees. And keep in mind that investment in MI can be exponential given that success from data insights often breeds the need for more data, sensors or meters to generate more data exhaust.
- The deepest insight can be generated from the most disparate sources
While many MI projects start with feeds from one division or government agency, the greatest insight occurs when intelligence is gathered across seemingly disparate units. For example, many smart city MI initiatives look for relationships between such unrelated departments as education and waste water; or streets and healthcare. Personal fitness devices increasingly serve as multi-nodal feeds across recreation, health, streets and housing. In the enterprise, one only need to study trends in social supply chain to see where some new dots are being connected with data integration.
So how is your enterprise or government organizing machine intelligence initiatives and where is the organizational center of gravity for these initiatives? What steps are you taking to prepare for MI 3.0?