Blog

Artificial Intelligence in business and manufacturing

Welcome to Business Futures podcast, the show where we take an honest and challenging look at the technologies and people that are shaping business. I'm Emma Pownall, Datel's Marketing Director - and in this episode I caught up with futurist Matt O’Neill and Datel’s R&D Director, Tim Purcell, to discuss the role of Artificial Intelligence within modern-day businesses.

AI has been a buzzword for well over a decade, and it’s slowly starting to fulfil its promise. Computers are already winning games of chess – and they’re starting to make operational decisions as effectively as humans in exciting sectors like healthcare and manufacturing.

So, what’s around the corner for manufacturers? What are the challenges? And how can we manage all the data that’s been collected by all these connected devices?

 

In this show, I explored:

  • How far we’ve progressed – and how far there is to go – in bringing Artificial Intelligence into the modern business

  • Why the Internet of Things is going to be AI’s best friend

  • The ethics of AI – and why you might be scared of smart toilets

 

Listen to the full episode here:

Join our mailing list to be notified on our next episode:

Your privacy is important to us.  We will only collect and use your personal data in ways that are described within our Privacy Notice, and in a manner that is consistent with our obligations and your rights under the law. Read Datel’s Privacy Notice for more information.

To begin with, I asked Matt and Tim for an introduction to what AI actually is...

Matt: Well, there's many definitions. But let's take the Wikipedia definition, which is very simply the science and engineering of making intelligent machines. What's quite important to remember about AI is that perhaps it's nothing new. You know, I would argue that it's got its roots going back as far as the 1940s, and 50s, particularly with Alan Turing, who came up with the Turing machine working with the lovely ladies at Bletchley Park. And that, of course, helped us lead the allies to winning the Second World War. But really, I think what we are in at the moment is this era of narrow intelligence or weak artificial intelligence. And this is where we've got clever algorithms working on faster processes in order to deliver very specific tasks. So we haven't quite got to that level of intelligence yet. But certainly, we are doing intelligent things.

Tim: So what we're saying is artificial intelligence really isn't quite as intelligent as we think it is. So really, it's just algorithmic programming, which is what we've had for quite a significant period of time.

Matt: Clearly, the algorithms perhaps on the whole, let's call them the mainstream algorithms, those that are being used in business today, they're not fundamentally that different to the 1950s. What is different is, of course, that they're running on very, very fast processors. And as of about 2012, they're being fuelled by lots of data, which is considered to be the fuel that is leading to the rise of the technology.

Tim: So technically, we've got a lot more power now today than we had previously. But actually the vast quantities of data that we're pulling in from different sources, different systems, and so on, it's that that's actually driving the technology now.

Matt: Absolutely, that's the difference maker.

I wanted to look at the role of AI within manufacturing.

Tim: It's probably having an impact across a variety of different areas, especially when you combine it with things like the Internet of Things. So by taking information from smart sensors, smart technology, actually, what we can now do is we can make predictive analytics, we can make decisions based upon that information. And AI gives us that facility to do that, which we haven't had previously.

Tim offered us a real-world example.

Tim: Well, if you take smart sensors in production equipment, for example. So production equipment now can feed in a variety of information such as its current stability, its current processing levels, and actually, the intelligence then could make decisions as to actually some of those parts might need replacing, we might be able to increase production based upon the information that we're being fed. So especially things like the automobile industry, where we're seeing cars built now with sensors in certain key components, so that some of the latest vehicles are able to detect when, for example, they need a service or when certain parts are degenerating, and they're able to then advise the driver, but also contact the service depot to say that actually the car needs to be brought in, and therefore arrange an inspection or arrange a service directly with the company itself. So we're starting to see that type of technology become generally available and not just limited to the automobile industry, we're seeing it across a number of different technologies as well. So I've recently read an article that Microsoft have produced around a lift manufacturer, where the lifts are able to detect whether their parts are failing, whether they're going to need some form of maintenance applying to them and actually contact the service engineer to come out and fix the lift before actually anybody notices that there’s a problem and therefore there's a disruption to service.

Matt: Yeah, I think there's another exciting area as well for manufacturers, which is how can you use all this data that's being generated by sensors, and apply it to the design process? You know, and where does AI start and a human designer stop? And it seems obvious to me that say, data driven products are entirely likely to become a reality and perhaps, for smaller and medium sized manufacturers, you know, taking that data and incorporating it into your next iteration can only be a good thing.

Tim: A good example as to where a business is actually using a certain level of intelligence is in conveyor tracking, for example. So being able to intelligently determine where a box should end up for picking by analysing where the order needs to go to, which pickers need to be given access to it, where the stock is within the building. And again, that's just fed by data. So having the visibility of what stock you've got in your system, where that stock resides, that is fed into the automation system. So the conveyor belt system is able to pick that up, identify which order a box belongs to, and then route it to the relevant part of the warehouse for the picker to feed that into. We're also seeing a lot of investments in automated guidance vehicles. So for example, forklift trucks are being guided by intelligence. So they're being guided based upon the information that's being fed such as where they need to go based upon the order profile that they're picking. We're also seeing at the end of manufacturing plants, as well as production is dropped out, that robots are able to understand what is being fed off, and therefore where those items need to be fed to. So the capacity of the pallet that they’re being dropped onto, for example. And then once a pallet has reached its capacity based upon the product that's being loaded onto it, when that pallet needs to be moved elsewhere in the warehouse. So we are seeing a lot of businesses today within our customer base that are actually using this type of technology. And the sensors that are generating the information to identify when a pallet is full and when a pallet needs to be moved are all being fed into some fairly complex intelligence engines that are then able to decide where to send that pallet next. So you're almost moving towards that completely lights-out type warehouse structure where actually an automation is guiding all of the production right from the moment it's dropped off the production line, locating it, and then moving into the relevant place in the warehouse store.

Matt: At the core of what Tim was talking about earlier, this idea of the Internet of Things are, he quite rightly points out, lots and lots of sensors. And these sensors can take many forms. You know, we've got motion detectors, we've got accelerometers which measure speed, we've got gyroscopic sensors, which measure angle of tilt, I think once I counted that in across a range of different smartphones, there were 14 different types of sensor, one of which, interestingly, after the Tohuku earthquake in 2011, Sharp introduced a Geiger counter into one of their mobile phones. So, powering this IoT is absolutely as Tim alluded to, a range of different sensors that can take data from our environment. But where I think it gets really interesting for business is introducing actuators, so actuators, this ability to change energy into motion. Yeah, and we see it even in our phones in the form of vibrational motors. So you know, at the core I think in manufacturing, IoT is really important. Where another point, I think that adds significant value for manufacturers, is of course, traceability, and that applies specifically in food production as well - being able to trace from farm to fork. So IoT most definitely is finding itself felt, I would say more significantly in manufacturing than perhaps in the domestic environment.

Tim: It’s probably also being fuelled by a number of other different technologies as well. Technology such as blockchain, for example. So where we're able to generate and feed information into different systems, such as a blockchain system for immutability, for example. I think that all feeds into that intelligence piece that we're having to use a certain level of sophistication then, to be able to interpret that data that's going into these different systems, and the vast amounts that we're actually generating as well.

I asked Tim if he thinks we're seeing any examples of this within our manufacturing customers that Datel work with.

Tim: I think what we're seeing is a number of customers taking the initiative to implement varying degrees of both artificial intelligence and sensor based systems. So what we're finding is that as businesses are trying to solve challenges and problems and increase productivity and efficiency, what they're doing is they're finding solutions to these problems by looking at the technologies that are available. So we are seeing some of them using things like Internet of Things technology, but also blockchain and also artificial intelligence to pull the analytics out of that.

It seems like this could potentially bring some challenges for businesses, and I wanted to know how customers are actually overcoming such challenges.

Tim: I think the biggest challenge actually with the vast amounts of data that we are producing is actually being able to interpret it. So whilst intelligence can give us a certain level of analysis, it's actually having the domain knowledge to be able to power that. So to be able to understand it, to be able to interpret what the information is being presented back to us is. And I think for me, that is probably the biggest challenge today that businesses might just ask for, can we have some form of business intelligence? Well, actually, what really is that? And it's having that knowledge to be able to understand to be able to spot the patterns to be able to spot the anomalies within the data. I think that's where the challenge is today.

Tim suggests that we are actually finding some solutions to these problems for our customers.

Tim: I think there are, there are solutions starting to present themselves. One of the biggest challenges is just the skillset as much as anything else, so I think as we're evolving as technologists and as businesses, what we're having to do is find the right skillsets to solve some of those problems. I think with the vast amounts of data, it's trying to find data specialists to be able to help support the businesses. I think that is a big challenge that I think the number of data scientists that are available to the community as it stands is very, very limited. But I think that is gradually starting to evolve as people are starting to understand what data is going to do for the business, they're starting to find ways of understanding how they can utilize that information. I think also some of the intelligence engines that are being built today: there are some fantastic examples out there for things like credit forecasting, stock forecasting, for example. So some of these technologies that are being built to solve this problem are starting to help as well. So we're able to feed our business data into this type of cloud-centric intelligence that is then able to at least start to build answers for us. Yes, we still have to take that and interpret it, but it's actually doing a lot of the legwork for us.

But how does this link into the business today? And what's the roadmap for a manufacturer thinking about embracing AI within their businesses? I asked Matt and Tim for their thoughts.

Tim: I don't feel that there is a single blueprint for this. So there is no one Technology Roadmap, I think for businesses, I think the answer could be industry or even company specific. To make AI work effectively, you're going to need data. And that's going to be generated in the form of sensors in equipment, intelligent devices used throughout the manufacturing processes, for example such as pallet tracking, such as monitoring conveyor systems, and the effectivity of conveyor systems. So typically, a business will ask the question as to how they can improve their relationship with customers and suppliers. So how do they improve their services? And once they've identified that, then they can start looking at how IoT and artificial intelligence may improve the business.

Matt: Can I just add something to that, I wonder if it also starts, for say, more ambitious or well-funded businesses especially, with working out what the processes of the business are. Because we have to establish what the processes are before you can then start to automate them of course, the big AI zeitgeist this year is around RPA, robotic process automation. But you can only automate things once you know what the process is you're trying to automate in the first place. So that seems to me to be a fairly good starting point, at least.

Tim: It's the age-old problem isn't it. It’s trying to identify where you as a business, maybe you're struggling or failing so if your relationship with your customers and suppliers is struggling, because you're not able to provide the right information to them, and so on. Absolutely right, it’s a case of trying to identify those processes that need some form of improvement, and then looking at how you can take the technologies of today, including IoT, including AI, and layer those on top of those processes. So for example, Siemens have embedded sensors in gas turbines now that monitor temperatures, stress, and other variables, and they feed that into their artificial intelligence engines. And then therefore, they can regulate fuel based upon the conditions of the turbines and outside variables, such as the weather.

Matt: And also I think, smaller manufacturers shouldn't be put off by this technology, you know, again, if we go back to the smartphone, one thing that the smartphone has bought is, is a plethora of cheap sensors, you know, if you've got an idea for something now, I don't think that you're limited by a lack of capital necessarily, you know, you can just hit up Alibaba, you obviously have to know what you're doing and what you're looking for. But the barriers to entry now in this stuff, I think, to at least take yourself to a prototype stage are no longer what they once were.

Tim: Absolutely, I mean, cost of technology now, certainly around smart sensors is coming down phenomenally. So the cost of maybe a GPS tracking device 10 years ago might have been 50-60 pounds per device, whereas now it's nothing short of pence. So being able to take such technology and embed it into systems and into processes now is just generally available. It's something that any business actually can take advantage of, and utilize that and also the data that that generates. We're also seeing maybe industries that we wouldn't expect technology to play a major part in actually starting to adopt this type of technology as well. And certainly businesses within locations that, again, may geographically have struggled to use technology. I think with the advent of things like 5G and the wide availability of networking, I think we're able to start utilizing this in different locations. We're seeing a lot of this and I suppose technically it’s a manufacturing industry, but in food processing, why farms in very, very remote locations are able to start capturing information against the crops that they're picking, things like the humidity, the temperature of those goods and able to track those all the way through effectively from the vine all the way through to the table.

Clearly, this is becoming an area that manufacturers can really tap into with costs going down and the availability of these sensors for businesses. But I wanted to know what practical steps these businesses can take to embrace this and potentially step forward and be someone at the forefront within their industry.

Tim: Ultimately, I think it comes down to what Matt was saying earlier on about identifying which processes need to be improved; trying to find places within the business where you can apply intelligence, where you can apply sensors, where you can apply automation, for example. And then taking that and understanding how those processes then might play out with that layer of intelligence applied to them. Ultimately, it comes down to data. So we need to feed the right data into the systems to take those intelligent decisions that we can apply automation to; we can feed into intelligence engines. So I think businesses are going to have to look at the type of information that they're working with, the type of data that they might be sharing, the type of data that they might be producing, or they could produce and then look at how they can generate that information by applying sensors and then putting an intelligence layer over the top of that.

Matt: That's all well and good. But I think that there's a strong ethical dimension that we need to be considering. For example, privacy. I mean, back in July, earlier this year, we saw Elon Musk announce the birth of his Neuralink company, this could be the birth by 2025 of neural interfaces, brain computer interfaces. And there's already work being done in neuroscience at the moment to investigate something called the P300 signal. And this is about measuring intent. So buying decisions, whether somebody has the inclination to buy a product, or even voting choices, so who gets to own our deepest data? And I think that that's a question that we need to be asking, in society. And then also control. You know, you mentioned algorithms earlier. And I think we're going to see this in all sorts of our household objects. You know, if we take the simple toilet, for example, people are already talking about the smart toilet. This is a toilet that's capable of measuring whether or not we might be prone to disease or whether we've been conforming to our diet, so that that has some positive impacts, perhaps for helping us with our health. But there's also a strong downside to that, because it's more than likely that as things become more connected, insurance companies might insist on having that feed made available to them. So, you know, perhaps we move from the smart to the surveillance toilet. And then I think alongside, technology is moving so quickly at the moment that perhaps progress is taking place farther than our government's ability to regulate it. And the challenge around that at the moment is that it's all being developed through purely the profit motive. And there's not enough caution, I think being given to whether or not these technologies could end up hurting us rather than aiding us.

Tim: Where do we draw the line, though? Where does surveillance actually go beyond surveillance and become almost snooping? So, if we take employees for example, companies are asking employees to wear smart devices, for example, to track biorhythmic detail, so to predict stress and so on, but actually, where do we start drawing the line? Because that starts to become snooping onto people's private information. And we can then start to predict sick patterns. And again, there is an ethical concern there that I'm not sure whether there is an answer to that at this moment in time.

Matt: The only answer I can give you is, I think we're figuring it out. You know, I mean, these developments are happening faster than our ability to regulate it and legislate for it. So, you know, technology seems to have all sorts of unintended social consequences. I don't really have a meaningful answer for you at the moment, Tim, but ask me tomorrow.

We discussed the ethics of business data, but what implications are there in terms of people's jobs? I asked Tim and Matt for their thoughts on the idea that robots could take over the manufacturing industry.

Tim: Since the dawn of the first job, people evolve, businesses evolve. And I think there is an ethical consideration for businesses to consider how staff might be displaced, and therefore, do we need to consider how to skill those staffs in different areas of the business? And I think naturally, Matt, probably, there will become a new generation of workers that are geared around intelligence and how to support the intelligence and the information that's being generated.

Matt: Absolutely. I mean, going back to what you said about robotics, I don't know if it's coming anytime soon. And it's for one very simple reason that, at the moment, robots are quite expensive. And there seems to be I think, a lot of short termism in most western businesses, you see a lot more incidentally, long termism in Chinese businesses. So yeah, for as long as human beings are more cost effective than robots, which that would appear to be for a long while, I think humans are fairly safe. The other thing I'd point out is that actually, if you dig into the data of robots, so if you look towards the International Federation of Robotics, what you'll find is that in the next two to three years, it's only forecast that something like, the number of multi-purpose robots in the entire world is about the equivalent, I think of 5% of the 200 million currently unemployed, right? So just by the numbers, they're not coming to take our jobs anytime soon. That's not to say they won't. And indeed, I was recently introduced to a business in London, Automata, who are introducing low cost robotics at the moment. So the situation may change, but the data in the short term isn't bearing that out.

Tim: And how's that different actually then to the evolution of production lines over the last 50 years, because they in themselves have displaced workers? That whereas production lines were historically always a very manually-intensive particular role, actually, the evolution of that within business has changed anyway. And actually, those people have been moved into other areas, other parts of the business, different businesses, for example.

Matt: But I think in production lines, yes, certainly, there's been a number of people displaced as a result. But those are single purpose robots. You know, those tend to be robots that perform a single function. What I'm alluding to here is multi-purpose robots or even we're starting to see the rise of so called cobots. So these are robots that we work alongside. Yes, jobs are being displaced. But that's nothing new. I mean, if we go back to before 1950, a common sight on the streets of industrial Britain would have been a “knocker upper”, you know, the knocker upper, the person who would walk around the streets waking you up. And why was that? Because we didn't have reliable alarm clocks. 1950 arrives. We have reliable alarm clocks, Mr. or Mrs. Knocker Upper is out of a job. Did they find another job? Of course they did. So I think one of the challenges that we've got going forward with more of this automation is that you know, how far up the value chain can everybody go? You know, it's fine if you're a skilled knowledge worker, you know, and you've got a piece of software that's augmenting a lower level, repetitive task. But if you're in a lower level menial job, and automation affects you, as it has proven to do since the beginning of the Industrial Revolution, then it could be more challenging, but one of the solutions that's being proposed, in fact, we only have to look across to the United States with Andrew Young, the presidential candidate who is suggesting that what we need as a society is universal basic income, the idea that everyone is given a fixed level of income in order to take care of themselves. And so that is a potential solution. So I think that society is having to figure out new ways of organising itself. And perhaps that means that we need to take a different attitude to what a job and work is. Perhaps many jobs will fall away. But with the universal basic income, people still want a sense of identity that perhaps comes from doing fulfilling work.

Alan Simpson, Datel’s Executive Chairman, invites you to share your thoughts on how AI could contribute to your business.

Alan: Over the years, AI has got more advanced. We've now got some actual practical products that we can provide to customers. One good example of this is it on the topic of fraud detection. I was reading a report the other day that in the US, a typical organisation loses 5% of its revenue to fraud each year, which is a frightening statistic. And probably it’s higher than that because people won't admit to it. But we now have products that we can supply to our customers that search through the masses of data they have in their ERP systems. And the software identifies patterns, and more importantly it identifies unusual or anomalous transactions that it then reports and recommends what the customer can do to improve their processes. So it's actually identifying fraudulent transactions and helping improve their businesses to overcome it with massive savings.

AI now is getting more realistic and affordable but how it relates to our customers that are using ERP systems, I have to be honest, and I'm not sure at the moment. What we would really like is for customers to come up with ideas, suggestions on how they think AI machine learning could work with their ERP systems and their ERP data. And then we are extremely keen to work with them on what we will call proof of concepts to understand just how they work and whether they can be done and whether they deliver actual benefits for the customer.

Looking forward, it's certainly exciting times we've got customers with masses of data in their ERP systems. If we can point one of these sophisticated AI tools at that data, we don't know what it's going to come up with. But I'm certain it will identify patterns, it will identify inefficiencies, and from that we can help our customers improve their processes and therefore become far more efficient.

My guests were futurist Matt O’Neill, Datel’s R&D Director Tim Purcell, and Datel’s Executive Chairman Alan Simpson. In this show, we’ve looked at AI – the opportunities, the first steps, and the ethical questions.

In the next show, my guest is Henry Rose Lee, an expert in intergenerational management. We’ll be looking at what the different generations want from their leaders, how managers can get the best out of employees with very different needs, and what technology can do to bring them all together.

See you next time on Business Futures.

 

This podcast was produced by ModComms, a full-service marketing agency offering innovative approaches to client challenges. www.modcommslimited.com 

My guests:

Matt O'Neill

Matt O'Neill is a futurist helping organisations prepare, plan and create their positive futures. Far from being a soothsayer staring into the misty distance, his forecasts are based in a mix of horizon scanning with a sprinkle of creativity. www.futuristmatt.com

Tim Purcell, R&D Director, Datel

Tim’s interest and passion for emerging technologies and how they can assist businesses to be more successful is at the heart of Datel’s research and development. Tim’s growing team is focused on keeping abreast of technology, how it can benefit businesses and how we can satisfy any gaps in business system requirements. Datel’s own product range, Fusion, is authored from within this team through collaborating with customers on their changing needs.

Alan Simpson, Executive Chairman, Datel

Alan Simpson founded Datel in 1981 when affordable business systems were a new concept and businesses needed help and guidance to get started with them. Since then the technology has changed but Datel’s approach hasn’t. Alan prides himself on building a consultative and collaborative culture within the business that enables the team to get close enough to our customers to build long term partnerships.

Business Futures host:

Emma Pownall, Marketing Director, Datel

My team and I provide our customers with a range of events, guides and tools that bridge the gap between business leaders and technology.  From large conferences connecting customers with each other and the software world, to sharing customer stories that explain what is possible with the right business solutions, I'm focused on sharing how people and technology can support business success.