There are few things in the business world as cringe-inducing as the question “what is your superpower?”.

I don’t have a superpower – I am a mere mortal. I do have a particular skill in taking quite complex business and technical concepts and explaining them in interesting and semi-amusing ways, but that does not make me Superman or Spiderman. It doesn’t even make me Aquaman and let’s face it, he’s pretty rubbish.

The last time I discussed superpowers before they entered the lexicon of annoying business jargon was when I was 16. One morning on the top deck of a cigarette smoke filled bus (yes I’m that old) my friends and I had completed our usual reciting of the most memorable lines from last night’s “Blackadder” and had moved onto the topic of what would be our desired superpower. Invisibility was a popular choice, largely used to avoid doing cross country runs, and to sneak into the staff room and listen in on our teachers’ conversations which we were convinced were about how much they hated us. One friend wanted Hulk-like strength so he could beat people up which, in hindsight, should have been more of a red flag that it seemed at the time.

One of the members of our friend group joined us on the top deck, looking disheveled even by our low standards, and when we asked him what his superpower would be, he replied instantly without asking for background information or context “being able to fly so I could fly to the bus stop in the morning.”

All of us looked perplexed, bemused, and confused, a not too uncommon look for us to be honest.

“Couldn’t you just fly all the way to school?” one of us asked, on behalf of the entire group.

“Oh yeah”, came the reply, “Suppose so.”

And this brings me on to AI.

AI is not magic. Nor is it a superpower. The very knowledgeable Matt Wood at AWS recently described it as “a mathematical parlor trick”. https://venturebeat.com/ai/aws-exec-downplays-existential-threat-of-ai-calls-it-a-mathematical-parlor-trick

The potential of AI, however, is huge. It goes way beyond the hype.

This isn’t 2023’s Metaverse (Alack, poor Metaverse – no one knew him, cared about him, used him, or wanted him, Horatio).

AI, both ‘traditional’ and generative, provides us with an opportunity to reimagine how we work, how we engage with customers, how we interact with suppliers and partners, how we…….

My former colleague, futurist, and all-around great guy Dave Aron, taught me one of the most useful lessons of my career – humble disrespect. We must constantly question why we do things, how we do things and how we can improve. We need to imagine the art of the possible and plot paths to potential futures.

We are at risk, at times, of just ‘doing AI’, just as in the past we were ‘doing digital transformation’ or ‘doing ERP’.

An Intelligent Enterprise leverages AI/ML embedded in its processes and systems to both improve and automate existing ways of working, but crucially also to innovate and expand its margin of differentiation.

The key is to always, as author Simon Sinek says, “start with Why?”. Why do our customers choose us over the competition? It isn’t and won’t be because you ‘do AI’. It is because of what AI enables you to be able to do for your customers.

AI is not the goal. It is what we can do with AI that will deliver value. ‘Doing AI’ is having the superpower of flight and flying to the bus stop to await the same bus to take you on the same journey to the same destination with the same experience every single day.

Matt Wood states that AI must be positioned to foster innovation. He states that any large shift is 50% technology and 50% culture. When it comes to changing how we work, how we differentiate to our customers, how we engage our employees, manage our suppliers, and improve our partnerships, then sometimes the balance changes to 10% technology and 90% culture.

Voltaire, the convivial social butterfly that he was, is quoted as saying:

“Our wretched species is so made that those who walk on the well-trodden path always throw stones at those who are showing a new road.”

Obviously, some deep-seated issues there, but he makes a good point. Change is hard. It always has been and always will be.

The world is buzzing with new opportunities and possibilities. But as Voltaire reminds us, you will never be able to show a new road if you always travel on the same bus. 

To find out more about how SAP can help you on your journey visit:

SAP.com/AI

or Chat with SAP

In the last article we described the HMI of an I4 system. There is one very important aspect of the HMI that deserves special attention: alarms.

In an industrial control system, alarms have a critical role in the safety and well-being of people and the facility. Alarms cannot function in the casual way that your alarm clock works at home. Alarms require careful engineering and a comprehensive system that alerts, records, and confirms acknowledgement. Alarms continue to alert until someone acknowledges they are aware of the condition. The alarm remains active, even though acknowledged, until the condition clears. The timestamp and pertinent information about the alarm condition enable diagnosis of the condition. Alarms can be prioritized and segregated by process area.

Unfortunately, alarms can become dangerous if implemented improperly. Some of you might recall an accident at a nuclear power plant called Three Mile Island in 1979. This accident became a bigger problem than it should have because a badly implemented alarm system caused a flood of alarms to occur when the condition began, which confused operators and obfuscated the true source of the problem. One of the control room operators stated “I would have liked to have thrown away the alarm panel. It wasn’t giving us any useful information.”

That is why ISA create the 18.2 standard for alarm management. It defines a methodology for creating and maintaining an effective alarm management system.

The process should begin when the project begins. Alarm design should not be an afterthought that is ignored until startup. An automation project needs to begin with alarms considered in the design. There must be a consistent philosophy to the alarm design that will make sense to operators. ISA 18.2 lays out the steps by addressing questions such as:

• What is considered an alarm?

• What is warning, severe, emergency?

• Will operators see the same indications in all situations, systems?

• What does an operator do when the alarm occurs? There should be no alarm configured without a course of correction.

There are cases, such as Three Mile Island, where a single fault can cause a cascade of related measurements to go into alarm. Conditional alarming can be used to avoid this condition. Logic can determine which signal is the root cause of an abnormal condition so that it is the only alarm that triggers. This allows an operator to focus on the root cause of an incident. This can also be used to disable alarms on idle equipment or shutdown process areas to take them off the alarm list when they only clutter up the display.

When the system is in operation, alarm management is not done. The alarm system must be routinely monitored to ensure it is performing properly. Evolving problems with alarm systems can be diagnosed with tools that can detect problems such as:

• Chattering: Alarms that repeatedly engage and annoy the operator, resulting in an environment that encourages alarms to be ignored

• Dormant: Alarms that are never cleared and therefore do not actually detect new alarms

Changes in alarm configuration need to be managed and documented. Ideally, a system should automatically log such changes.

I have seen a lot of alarm systems in my career with every alarm configured for emergency priority, continually blinking, and being ignored by operators. This is the classic boy cried wolf problem; when a multitude of alarms go off with no apparent impact, a critical alarm in the future will be ignored which could be very dangerous.

In the I4 era, we are fully connected, and alarms could come from a lot of places. Implementing alarms in a control system can be done in several different ways. Alarm could be implemented at Level 1 and displayed at Level 2, or implemented at Level 2 in a common gateway, or implemented in Level 2 at each HMI display. There are advantages and caveats to each approach. 

While ISA 18.2 gives a great methodology, and there are helpful tools to monitor for problems in an alarm system, alarm design is tedious work with no real shortcuts. However, the cost of doing it badly is too high to take the easy way out.

In these columns, I often see my role as being one of playing off the two worthy columns above me here, hence the title this month, "AI and Alarms." And, no, I am not alarmed by AI.

And I don't think AI is about to take over the world. Show me an AI driven robot that can change a bed or fold a fitted sheet, and I'll be worried. It is more likely that someone will sell me an alarm that says my sheets needed changing (bacteria count) than a robot that can actually change a sheet.

In reality, you have been using a form of AI for quite some time. That piece of software that finishes your sentences or fills in a word in your writing is AI, you have just never called it that.

I have had an AI device around the house here for five years. It is a Husqvarna Automower. It mows the grass and knows when the grass has been mowed enough. It works in an entirely random manner, but knows where it has been. When it is in an upset condition, it sends me a message on my phone informing me it needs help (it can't do everything).

It will be obsolesced at the end of this season, for you see, it needs a boundary wire in order to stay within the yard. New ones now do not need boundary wires, their GPS signal receivers are such that, after one pass with a human guiding them, they know where to go.

I had an oncologist tell me five years ago that it was even then, impossible for an oncology researcher to keep up with the technology without AI. As he put it, if a researcher was up to date on all the latest developments when they came to work on any given morning and diligently read all day, they would be five years behind by the end of the day. Only AI could keep up with the developments.

But what I really worry about with AI is that it will engender laziness, especially among those that should be writing for a living. And no, I am not talking about journalists per se, I am talking about technical professionals charged with writing instructions, standards and other such material. It seems like it will be a long time before AI can capture all the information required here on a stand alone basis, for much of this is basic, foundational information that will not come from the world wide web.

The human mind is not out of business--yet.

IoT inventory control technology can transform facility management, increasing visibility, and efficiency while reducing waste. Manufacturers can simplify item traceability and inventory monitoring using innovative IoT and RFID technology. What are the potential benefits of adopting IoT inventory management in manufacturing? 

In today’s advanced industrial era, effective maintenance is paramount. Companies grapple with the repercussions of inadequate maintenance strategies, leading to unexpected downtimes with staggering costs of up to $260,000 per hour.

It’s no secret that the manufacturing industry is undergoing a massive transformation at the moment. To stay competitive, manufacturing companies are under pressure to implement smart manufacturing solutions that digitize and automate processes and workflows.

AI Takeoff refers to a specific inflection point where Artificial General Intelligence (AGI) – AI that can perform any intellectual task a human can – reaches a stage of recursive self-improvement. This would mean a sudden exponential increase in intelligence, often referred to as an “intelligence explosion,” a phenomenon that would irreversibly alter the course of history.