Johannesburg, 21 Nov 2019
The Internet of Things (IOT) is the magic dust of the digital age. By sprinkling sensors and edge devices across a site, the physical and digital worlds converge through streams of data. It is not even a complicated arrangement: sensors are narrowly focused and tend to do one job. As their prices have fallen, it has become feasible to increase the number of sensors and read even more from an environment.
This increases access to relevant data, something businesses have been able to put to good use. According to the International Data Corporation’s (IDC's) "7th Annual Global IOT Decision-Maker" survey, released earlier in 2019, satisfaction levels with IOT projects are high. Not only that, but a failing IOT project revealed itself very quickly, within months.
So, the price of failure is low, and the chances of recovering for success are good. No wonder 85% of the respondents had IOT budgets already in place.
But there is still a problem, and a pretty big one at that.
New data, old tools
Accounting firm Deloitte also released an IOT survey this year, focusing on the relationship between IOT and artificial intelligence (AI). It echoed the good vibrations around IOT as well, yet made a startling revelation: 70% of respondents use spreadsheets to process IOT data.
We can credit the spreadsheet for opening business’s eyes to the original potential of computers and digital technology. But it’s woefully inadequate for managing the complex and large data sets being generated by numerous IOT devices. Even if a spreadsheet could handle the job at first, they fall behind as IOT projects scale. Such a manual intervention also cancels out any automation strategies.
IOT requires more than a spreadsheet. To match its quality, volume and speed of data, it fits best with AI.
“While AI mimics human behaviour through its intelligence systems, IOT has no intelligence,” says James Ndegwa, Architecture Lead Manager at Westcon-Comstor. “Its key role in technology is to complement the IOT systems towards services for the business.”
As mentioned earlier, IOT sensors are pretty singular devices. The wider IOT system can also consist of more sophisticated IOT devices, such as gateways. Such devices help clean and share data, but they don’t analyse it.
That task is still often left to humans, which doesn’t make much sense if you hope to automate a system. For example, says Ndegwa, it could hold back routine but important tasks such as inventory management: “A store can put IOT technologies in its shelves or through its cameras to monitor what happens to stock. But if the resulting data is processed manually by a person, it defeats the purpose.
"It creates a new bottleneck that can’t keep up as the IOT footprint grows. AI can automate those processes. The AI can process the IOT data and then inform the inventory system.”
IOT hearts AI
It might scare some to leave the interpretation of operational data to a machine, but AI is not a homogeneous description. Instead, it encapsulates a broad spectrum of different AI systems. Some are more successful than others, yet the vast majority are trained to deal with very specific tasks.
“There are different types of AI, and the differences come in through context and the industry where the AI is being used,” says Ndegwa. “You should look at AI as a very specialised worker that only does one job. Once it’s outside of that context, it is not of much use.”
This also means that the AI’s purpose and responsibilities are entirely within the control of the business. In a sense, AI is perfect for IOT: both are narrow in their purposes, both handle routine tasks, and both exist to create output that serves the company.
So, if a fear of AI is your concern, you are making a big mistake with regard to your IOT investments. It limits your ability to automate parts of your environment and get better value from the data being generated. If that’s in doubt, a staggering 99% of the Deloitte survey respondents said AI+IOT met or exceeded their expectations.
But isn’t creating an AI-automated environment challenging and expensive to implement? Ndegwa dismisses the first problem, noting the ample experience the market has accrued around AI solutions: “It isn’t difficult as this has been done before. The key is to identify what you are trying to solve and feeding the relevant information that AI needs to work on. The main challenge of AI is usually contextual as it relies on the data given and not on any unique situations.”
Cost is a bigger consideration, but here he urges companies to think about the investment: “AI is an investment to an organisation. It is resource-intensive but has a great return on investment to it once used in the right way. It can pay for itself.”
In tandem to this, a good appreciation of AI in an organisation goes hand-in-hand with good data practices and use of analytics. Both the IDC and Deloitte surveys reinforce that relationship: companies that can make data their own can flourish through AI.
But, Ndegwa added, thinking of the business outcome is imperative because every situation is unique.
If you've ever wondered if there is a place for AI in your company, start with your IOT data. These are often routine and simpler data sets, attuned to operate within specified limits. An AI is a perfect companion to nitpick the data and make intelligent decisions on your behalf.
Or you can spend another evening on a spreadsheet as the IOT data mountain grows behind you.
Share