FAQs (frequently asked questions) on use of artificial intelligence (AI) in supply chain management
Following are some of the frequently asked questions (FAQs) on supply chain artificial intelligence that we have encountered in our speeches, workshops, seminars, and other forums.
Feel free to ask more questions if your particular question is not answered below.
If you have read our FAQs on supply chain analytics, you would have read that a number of supply chain analytics tasks can now be replaced by artificial intelligence algorithms. Whether we talk about the descriptive supply chain analytics, or diagnostics supply chain analytics, or predictive / prescriptive supply chain analytics, there are a plethora of situations where human intelligence can now be gradually replaced by machine intelligence.
Let us take the same example that we used in the FAQs on supply chain analytics to talk about supply chain AI.
Let us say that your supply chain descriptive analytics determined that on Sunday the truck productivity is 17% higher due to lighter traffic. A human would use this information to conduct further supply chain analytics to determine whether it will be worthwhile putting extra trucks on the road in response to this information.
Having made that decision, as the traffic on the roads gets higher, the Sunday advantage would start to disappear. Where is the cross-over point when there is no advantage to running extra trucks on Sunday? Can AI build an analytical tool drawing real time truck transit time data from a GPS tracker to determine the point where the traffic has become heavy enough to stop the extra trucks?
This is just one potential application of AI combined with IoT and supply chain analytics to make smart real time decisions. There are thousands of such potential decisions in every supply chain. Some are worth investing into, while others may not have these kind of rick returns.
For a human to be switched on 24/7 and make real time decisions as soon as the data becomes available would be very costly, if not impossible. We deploy humans in shifts when this real time decision making capability is absolutely critical - e.g. plants that are running on continuous flow processes such as oil refinery, merchant ships, glass manufacturing and many others. In most other situations we wait for humans to make these decisions in batches.
As the use of AI expands many of these decisions can be made by the machines following the algorithms built into smart decisions. Machines make these decisions in real time, cutting out the delays, thereby making the supply chain processes a lot more smooth, efficient and continuous flow. Bottlenecks and wastes are cut. Bull whip effect is reduced as a result. Supply chain planning and control becomes a lot easier. Cost of operations goes down, while the productivity goes up.
These advantages are in addition to the cost savings from having a machine replace a human to make the decision.
Almost all the current applications of AI in SCM are in the following three areas:
- Supply chain automation: after automation of the factories now it is the turn of the warehouses and eventually the trucks to give way to robotics. Although the current applications are still mostly in the experimental stages, the time is not far away when a large part of logistics execution task will be carried out by a combination of AI, IoT and automation.
- Supply chain decision support: Gradually, over the period of last 2 decades the algorithms embedded into the DSS are getting more sophisticated and accurate. As the crunching power, bandwidth and connectivity expands, and as data collection and availability become ubiquitous, DSS have access to far more accurate and complete dataset than ever before in the past. Sophisticated AI is now built so deeply into the supply chain decision making process that we just take it for granted, and do not even call it AI in many cases.
- Supply chain reporting: Almost every supply chain dashboard today deploys some sort of minor AI to make value judgement on the supply chain performance. Many others are also making a judgment on what data to report on, and what to ignore. Many others are going a step ahead and making recommendations on actions emanating from the data analytics - going into the realm of decision support that was earlier out of the purview of dashboards.
Whether they are replacing the human intervention, or supplementing it AI applications in supply chain are only going to grow. Almost in every case the value is accruing more to the AI vendors, than to the end user for the simple reason that the end users are neither initiating the projects, not participating in the value building stage of the AI smile curve.
Technological and social trends are colliding to create a perfect storm for development of AI in supply chains. COVID-19 has made remote work a norm. With it comes a widespread adoption of remote management capability and IoT.
On the technology front - automation, AI and IoT aid each other and create a flywheel effect helping adoption of each of them simultaneously.
Sure, companies will struggle for a period of time with sub-par applications - but if they make measured investments and keep experimenting, eventually they will gain a upper hand on the technology, and on their rivals. That is the secret to Amazon's current success after all.
Digital supply chain twin and Augmented reality (AR) are the current experimental models that are under development for aiding the supply chain management using artificial intelligence.
Advanced analytics, IoT and Robotic process automation (RPA) are the most mature applications in supply chain AI.
Advanced analytics is the autonomous examination of data or content using advanced techniques and tools to attain more drilled down insights, forecast or generate solutions. Advanced analytics are employed in real-time areas of supply chain management to make real-time pricing, quality assurance and updated replenishment possible.
The answer to this question depends on two things:
- Timing of application - are we talking about current applications of supply chain AI, or future applications of AI
- How to define most potential application - by number, or by business impact?
Let me explain. If we are talking about what is happening currently, as you have already read in the FAQs on supply chain analytics, current applications of AI are mostly in the areas of diagnostic and predictive analytics. The reason is that BIG data is still taking off and prescriptive analytics is too complicated to deploy AI on full fledged basis. But at some point in future AI will be equally deployed in prescriptive and descriptive analytics, once big data becomes the norm in supply chin analytics. When is that likely to happen? at the current rate of adoption it could take ten years but COVID-19 has actually accelerated the rate of deployment.
Similarly, in terms of number of application diagnostic analytics will win hands down for the simple reason that you cannot imagine a dashboard without some sort of AI capability built into it. However, in terms of potential business impact prescriptive analytics provides the biggest bang for the buck.
Planning required an ability to predict - however imperfectly. Supply chain planning incorporates predicting demand, supply and inventories.
Predictive nature of the AI algorithms enables the experts in the supply chain fields to forecast these numbers and plan accordingly. The higher the accuracy of predictability, the more robust the supply chain planning process, the better the business outcomes.
According to McKinsey, 61% of executives report decreased costs, and 53% report increased revenues as a direct result of implementing AI in the supply chain.
We, at GLOBAL SUPPLY CHAIN GROUP, are currently engaged in a project modelling the cost reduction impact of AI on supply chain. We will come back and report it here once the modelling project is completed.
The COVID-19 epidemic has resulted in the breakdown of supply chains globally, the economic impact of which will linger for years or maybe decades to come.
A recent survey by the Institute For Supply Chain Management, nearly 75% of companies reported some sort of supply chain issues dues to coronavirus-related transportation restrictions. What makes it worse is that the full downstream impact of the disruption cannot be known in advance. By using smarter technologies like AI a huge amount of data can be processed and such market anomalies can be predicted and a solution can be found out from data obtained during the crisis.
The most fundamental requirement is that the decision makers have a good understanding of supply chain management itself. The five flows of supply chains, the two key requirements of supply chains and four underlying pillars of supply chain management should be intuitively understood.
At the most basic level people should not confuse supply chain and logistics, or supply chain and procurement.
Once optimisation and integration - the two key requirements of supply chains become paramount goals - people will automatically get out of satisficer mindset and look for AI as the solution for optimisation.
With this understanding comes a willingness to take calculated bets on new technologies and make these bets pay off. And with that comes an ability to expand the winning technology bets and contract those that do not pay off adequately.
Finally, availability of data set, and willingness to collaborate and invest is needed to incorporate the use of AI into supply chain. The requirements of an organisation to implement AI includes High computing capacity, high data storage capacity, efficient networking infrastructure and security over network and data.
There is a lot of talk about digital supply chain twin.
A digital twin is a digital representation of a real-world supply chain environment. it showcases the relationships between all the relevant nodes of the supply chain — such as products, customers, markets, warehouses, manufacturing plants.
But this is mere talk.
At GLOBAL SUPPLY CHAIN GROUP, we work with companies that are obsessed about customer delight and creating minutest amount of additional customer value. None of these companies have shown much interest in creating a high cost digital twin with indeterminate customer value.
These are the companies which are investigating the possibilities of using robotics for customer deliveries and predictive shipping methodologies to further enhance their competitive edge.
The last mile problem of supply chain networks may eventually be solved by using AI as well.
Almost every industry will benefit from AI-powered SCM - but bigger the bets, bigger the potential gains from AI. Healthcare, upstream oil and gas industries, last mile deliveries and chemicals are some of the industries where we see big bets on supply chain AI at the moment.
Almost every ERP tool and every DSS tool wants to build the next clever system that will solve the world hunger. Afterall that is part of their overall mission. Some of them will eventually succeed in this quest. Even those who do not succeed with open up new pathways for the others. The role of the current ERP and DSS is to do just that.
Notes on FAQs
Clearly, any such list of frequently asked questions (FAQs) about supply chain can never be fully exhaustive. Neither is anyone, including us, the final authority and arbitrator on this or any other topic.
You will have your own opinions on many of these topics, and will have many other questions.
We throw open the comments section to you for your opinions and questions. We will try to address all of these, and the best ones will attract a reward in the form of one of our books, or publications.