Artificial Intelligence (AI) is human intelligence exhibited by machines and systems that approximate, mimic, replicate and automate, and eventually enhance, human thinking. Machine Learning (ML) is AI’s technological twin.
In this paper we will explore the ways AI and ML are affecting supply chain logistics today and its potential to innovate and disrupt it in the future.
Forward thinking companies are already harnessing the power of AI and ML to execute more strategic plays such as carrier selection and warehouse location as well as to enhance real-time decision-making on more tactical issues of inventories, vehicles, personnel and costs.
Today, there is more information to sift through but there are also more powerful computers that are able to sort, evaluate and accelerate the development and sharing of knowledge and corresponding action. The modern and future workforce is going to have to deal with this reality. As one global retailer stated in a recruitment ad- “we’re becoming more intelligent, automated and algorithmic…constantly re-imagining how we get the right product to our customers better, faster and more cost-effectively.”
In the supply chain logistics and transportation space, AI, ML and related technologies should allow companies to achieve greater optimisation and responsiveness end-to-end.
Here are but a few examples:
- Augmented decision making: Logisticians are asked to make choices involving large amounts of data. Consider choosing a trucking company and having to sift through an almost limitless pool of candidates - their routes, schedules, pricing, safety record, etc. Personnel using AI can automate the process and within a matter of minutes, narrow down options and let humans add intuition and non-quantitative factors.
- Predictive analytics: AI or more precisely a subset namely telematics and sensors (the Internet of Things) can pinpoint when a vehicle needs preventive maintenance or when temperature within a refrigerated conveyance is outside set parameters thus reducing the potential for breakdowns and cargo damage.
- A global cargo integrator (carrier and intermediary) developed a machine learning based tool analysing 58 parameters in order to predict the average daily transit time for a given lane. Additionally, it can also identify other factors such as climate and operational variables that can affect shipment delays. This same company also monitors millions of online and social media posts to identify potential materials shortages, access issues and supplier status all crucial elements that could presage supply chain problems.
- Strategic Optimisation: Supply chain professionals can utilise AI, ML and associated technologies to present a range of scenarios, such as deployment of inventories and transportation assets, enabling more informed longer-term Where, When and How choices. AI can generate Big, Clean (taking incomplete data sets and making precise deductions based on history along with Natural Language Processing to decipher and streamline foreign language) Data to optimise performance- think a small package delivery company saving millions of gallons of fuel per year by organising efficient routing, right turns only, of last mile segments.
- An offshoot of AI and ML, Natural Language Processing is being used in the freight brokerage business taking and structuring information from blast emails and conversations transmitted between trucking companies and brokers to identify available “smart” capacity in a tight market. This streamlines the time and labour-intensive load matching and reduces the natural time delay between listing of a load and finding a suitable and willing transport partner.
Other areas where AI is actively engaged are robotics, computer vision and autonomous vehicles.
Robotics is already embedded inside the supply chain. One research firm estimates that by 2021 worldwide sales of robots for warehousing and logistics will exceed US$22 billion. They are used to locate, pick, count, track and move inventory inside storage facilities as well as sort and convey oversized packages at ground transport distribution hubs.
Computer vision powered by AI can improve cargo or equipment inspection processes with the ability to quickly identify and classify damage and determine the appropriate corrective action. IBM Watson has been programmed with visual recognition capabilities to assess damage and wear to railcars. Cameras installed along train tracks feed images that can be quickly gathered and processed. Amazon is also using this technology to facilitate the unloading of trailers- cutting the time down significantly.
While driverless trucks may not be in the very near future, tools spawned from this concept are available today - highway autopilot, lane assist and assisted braking are stepping stones to true autonomy as well as making road transport safer which should result in fewer cargo claims due to road accidents - collisions, overturns and the like. Computer controlled driving systems also allow for multiple trucks to drive in formation to lower fuel consumption.
More mundane perhaps but already a growing part in warehouse operations are Autonomous Guided Vehicles or AGVs which use AI-based navigation and collaboration to learn and improve material handling performance.
Then there is the case for autonomous vessels that have garnered a lot of press but nothing quite tangible.
More Artificial Intelligence start-ups catering to the logistics and related industries are cropping up with no signs of waning popularity. AI is still just scratching the surface of capability even if its marketing is fairly mature.
AI and Machine Learning will become more prevalent and practitioners will find ever growing ways to use it. A key challenge is that people do not trust it; a result of the hype from media and AI “solution” vendors. Every function mentioned above falls in the category of Narrow Intelligence having a restricted scope dedicated to assist with or take over specific tasks. The next iteration and one that elicits fear from the public is General Intelligence which would represent systems with capabilities equivalent to human intelligence currently is non-existent in industrial applications.
Depending on the needs of the various stakeholders, the potential for these new forms of intelligence could focus on identifying internal process inefficiencies; addressing quality control; improving both the quality and speed of planning; mitigating disruptions from external events; or, moving robot-assisted activities forward.
Companies, including our assureds, will use, either directly or indirectly through their transportation carriers or third-party logistics providers, AI, ML and other cutting-edge technology. However, until truly autonomous trucks or ships become reality, exposures to loss and damage will remain the same. Unfortunately inherent in any automated system is the threat of cyber interference and one need not use much imagination to see how a security breach could wreak havoc.
Predictive analytics appears to offer the more immediate benefit to insurance and risk management, but the challenge is how to harness its potential in a meaningful, actionable way. Can, for example, existing loss data translate into reliable forecasts on an account, book of business or industry basis?
Ø “Artificial Intelligence in Logistics”: DHL, IBM and Singapore Management University, 2018
Ø “How Artificial Intelligence and Machine Learning are Revolutionizing Logistics, Supply Chain and Transportation”: Forbes Insights. September 4, 2018