Volume 5 Issue 9 September 2023

A
Paperitalo
Publication
In this Issue

Welcome to Industree 4.0 for September, 2023, exclusively sponsored by SAP.

SAP

By Thiago Prado,

SAP

Leveraging Artificial Intelligence in ERP

Artificial intelligence (AI) represents one of the most transformative advancements of our time, driven by a blend of science, technology, and imagination. At its core, AI refers to the simulation of human intelligence processes by machines, particularly computer systems. This simulation includes learning (the acquisition of information and rules for using the information), reasoning (using rules to reach approximate or definite conclusions), and self-correction. 


Definition and Brief History 


AI traces its roots back to the mid-20th century, with the seminal 1956 Dartmouth conference often hailed as the birthplace of AI as a field of study. Over the decades, AI has grown from a theoretical concept into a reality, and it now plays a key role in various facets of modern society. 


The evolution of Artificial Intelligence has given rise to diverse applications across industries – including the paper and packaging industries. In paper and packaging, AI can help to ensure manufacturing accuracy, reduce inventory carrying costs and enable companies to meet delivery deadlines. AI can also help optimize resource orchestration by automatically dispatching manufacturing operations to available resources. For partner industries such as transportation, AI powers self-driving technology, enabling vehicles to interpret sensory data and navigate complex environments. 


Artificial Intelligence in Enterprise Resource Planning (ERP) 


Artificial Intelligence (AI) is dramatically changing the landscape of Enterprise Resource Planning (ERP), adding a new dimension of intelligence to business management applications. The fusion of AI and ERP is revolutionizing how businesses plan, control, and manage their resources. 


AI in ERP refers to the integration of AI technologies into ERP systems to enhance their capabilities and improve business outcomes. By embedding intelligent algorithms into ERP software, businesses can automate processes, make more informed decisions, and uncover valuable insights from their data. This represents a significant leap from traditional ERP systems, opening up new possibilities for business optimization. 


Examples of AI in ERP 


  • Intelligent Automation AI enables intelligent automation in ERP systems, automating repetitive tasks, and streamlining processes. Whether it’s order processing, invoice matching, or data entry, AI can increase efficiency and reduce errors, freeing up staff for higher-value tasks. 
  • Data Analytics and Predictions AI can analyze vast amounts of ERP data to derive actionable insights. It can predict trends, identify anomalies, and provide recommendations to guide decision-making. This predictive capability is particularly valuable in areas like demand forecasting, inventory management, and financial planning. 
  • Customer Service and CRM AI is transforming Customer Relationship Management (CRM) within ERP systems. AI-powered chatbots can handle customer inquiries, while AI-driven analytics can provide personalized customer insights, enhancing the customer experience and improving customer retention. 


Benefits of AI in ERP 


Integrating AI into ERP systems brings numerous benefits. Intelligent automation can significantly improve operational efficiency, while AI-driven analytics can enhance decision-making. Moreover, AI can enable more personalized customer experiences and improve forecasting accuracy, driving business performance. By harnessing AI, businesses can extract more value from their ERP systems and gain a competitive edge. 


Challenges of AI in ERP 


Despite the potential benefits, integrating AI into ERP systems poses challenges. Data privacy and security are significant concerns. Businesses also need the right skills and resources to implement and manage AI-enabled ERP systems. Furthermore, the potential job displacement due to automation raises ethical and social issues that businesses must consider. 


The Role of Machine Learning in AI ERP 


Machine Learning (ML), a key component of AI, plays a pivotal role in AI-enabled ERP. ML algorithms can learn from past ERP data, refine their models, and make accurate predictions. Whether it’s predicting demand, optimizing inventory, or personalizing customer interactions, machine learning adds a layer of intelligence to ERP systems that can drive significant business improvements. 


Governance and Ethics of AI in ERP 


The implementation of AI in ERP requires robust governance and a strong focus on ethics. Businesses must ensure that their use of AI respects data privacy, maintains transparency, and considers the potential social implications. By addressing these issues proactively, businesses can ensure that their use of AI in ERP is both responsible and beneficial. 


Governance of Artificial Intelligence 


The rise of artificial intelligence (AI) has brought forth a new set of challenges and considerations, among which governance stands out as a critical factor. 


AI governance refers to the framework and policies implemented to ensure the responsible and ethical use of AI technologies. This includes considerations around data privacy, transparency, accountability, and security. The goal of AI governance is to ensure that AI technologies are used in a way that is fair, safe, and beneficial for all stakeholders. Within ERP systems, governance is important to ensure data privacy, maintain system integrity, and mitigate the risks associated with automation. 


Current Best Practices 


Best practices in AI governance involve establishing a robust governance framework that addresses key ethical principles. This includes ensuring transparency in AI decision-making, implementing strong data privacy measures, and maintaining system integrity and security. 


Furthermore, it’s essential to continually monitor and update governance policies as AI technologies evolve. Regular audits and reviews can help identify potential issues and ensure compliance with governance policies. 


Lastly, businesses should foster a culture of responsible AI use. This involves educating employees about ethical AI practices, promoting transparency in AI operations, and encouraging accountability at all levels of the organization. By following these best practices, businesses can harness the benefits of AI while mitigating potential risks. 


Ethics of Artificial Intelligence 


As AI technologies continue to transform business operations, ethical considerations are becoming increasingly crucial. Balancing the pursuit of innovation with the commitment to ethical principles is a challenge that must be met head-on. 


Ethics in AI involves addressing questions about the moral implications of AI technologies. These questions touch on issues of fairness, transparency, accountability, and privacy. With AI systems making decisions that can significantly impact individuals and society, it’s crucial to ensure these decisions are made ethically. A commitment to ethical AI use is not just a moral imperative, but also a key to sustainable and responsible innovation. The successful integration of AI in business depends on the alignment of AI applications with core ethical principles. 


Artificial Intelligence (AI) – United States Department of State


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




Disclaimer:

SAP notes that posts about potential uses of generative AI and large language models are merely the individual poster’s ideas and opinions, and do not represent SAP’s official position or future development roadmap. SAP has no legal obligation or other commitment to pursue any course of business, or develop or release any functionality, mentioned in any post or related content on this website.

Building I4: Level 2, Machine Learning

By Pat Dixon, PE, PMP


Vice President of Automation, Pulmac Systems International (pulmac.com)


and


Haoxiang Yu, Ph.D. Student; Affiliation:

Department of Electrical and Computer Engineering, The University of Texas at Austin



My July 2019 article introduced machine learning (ML) to this audience. It now time to put ML in the proper I4 context and elaborate on what it means to building an I4 system.


At the time of the prior article, ML was one of the many terms that proliferate in industry without clear definition. It still is ill defined. Therefore, there needs to be elaboration on this definition:


MACHINE: Non-human. The result comes from a machine (software, hardware), not a human being.


LEARNING: This is a more difficult term. What is learning?


If you type a bunch of numbers into Excel and tell it to perform a linear regression, the result comes from a machine. The regression algorithm uses the data to give you model coefficients and can predict an output value based on the inputs you give it. Did the machine learn?  


In one context, the machine changed from an unlearned state to a learned state. After it did its work, it now knows the model. In the same way that you went to school and learned things, and at this point in your career your retained 5% of what you were taught, you did learn from that experience. Therefore, it can be argued the machine learned.


For many practitioners, this is an unsatisfactory definition. Humans continually learn. We have the 5 senses such as sight, smell, and touch. We are always gathering data and our brains are constantly adapting. Whatever we were taught in school has probably adapted greatly as we experienced the reality of industry and life.


By human standards, Excel did not learn. Our standards matter because if they didn’t, whose would? Regardless of the multitudes of ways in which machines far eclipse our abilities to process and store data, we are the ultimate arbiters of what is considered learning. We are the masters who will determine whether the slave machines are smart.


The result is that for the purpose of ML, our definition of learning must be adaptive. To make machine learning effective, it needs to be organized as a system rather than a static model. This is because our definition of learning requires adaptability. The system must consistently gather data, refine existing models, and potentially create new ones.

However, there is a big problem here. Data can be noisy. Noise is any attribute of data that is not useful or misleading, and if used can lead to learning exactly the wrong thing. 


It turns out humans are really good at dealing with noisy data. In “Algorithms to Live By” by Brian Christian and Tom Griffith, human beings come out on top as the best learning machines in the world. Our senses and brains are ideal for pre-processing data so that we can filter out (discard) what doesn’t make sense or isn’t worth considering. Similarly, a machine learning model should capture the general trend and filter out the noise. We are not perfect, but we are better at it than any machine. When machines get data, how do they know what noise is? If the data is pre-processed by a human before it is typed into Excel, the machine doesn’t need to know about noise. In machine learning, how does a machine pre-process data in realtime?


Paper industry applications involve processes like recovery boilers. Other industries have lots of other processes that can go boom. Applying our definition of ML without understanding caveats is really dangerous. It is important that ML applications for industry are robust, with full appreciation for stability and insensitive to noise.


In the I4 era, we have data coming in from everywhere because we are connected in ways that didn’t exist in I3. While there are applications that are marketed as “ML” living in controllers at Level 1, most practical applications of ML live at Level 2 and above. Instead of a proliferation of gateways and protocols to gather the data, the Unified Name Space (UNS) concept is fundamental to making ML happen. Today MQTT with Sparkplug B is helping make UNS a reality.


A useful ML system, which must be adaptive as we defined it, is in its infancy. Many who claim to have deployed ML in industry haven’t. There will certainly be a growing role for ML, but when you are building your I4 system be very cautious and skeptical when you hear ML. You don’t want to learn the hard way.

AI and Root Data

AI is getting a lot of ink here these days. Not surprising. I have lived through many transformative technologies. Back when I started, for instance, there were no desktop electronic calculators. That was 1970. Many of you younger readers may not even be able to imagine what has come along since then.


AI, however, is a bit different. It is not just a dumb device dependent on the human operator. Its intent is to supplement or replace human thinking.


Organizations are rushing towards this because AI does something many can not do: be creative when it comes to difficult or thorny problems.


But here in lies the danger. For a long time, we have said, when it comes to computers: GIGO--garbage in, garbage out. AI does not eliminate this flaw. If the Root Data, as I am calling it, that is the source information upon which AI is dependent to exercise its creativity, is flawed or missing vital points, the creativity of AI will be no better than an unimaginative human, in fact may actually be dangerous.


There will be reports, sooner or later, that AI produced a flawed outcome. This will take a bit of the bloom off the rose and we'll find out that AI was not the ultimate solution we thought it would be.


AI will still be extremely valuable, and we will learn lessons as to how to make AI better, but nirvana it will not be. At least not yet.


Six Digital Transformation KPIs to Help Measure Success

By Eric Whitley

Digital transformation requires rigorous and targeted oversight to achieve strategic goals while measuring the return on investment. Yet, manufacturers select metrics that often provide little insight, measuring technology uptake rather than the transformation's results.

Read the full article here

The Rise of Connected Devices: Redefining the Internet of Things

By Itamar Kunik

The world of technology is in a constant state of flux, and the Internet of Things (IoT) is no exception. Today, “IoT” no longer refers to simple machines “just talking” to each other as it once was. Rather, modern IoT devices, or “Connected Devices,” are much smarter.


Read the full article here

The 'Whole & Parts' Solution to Securely Managing IoT Security Challenges

By Shiva Nathan

A complete IoT solution is a lot like a mini-internet — diverse devices spread across the world, connecting via various network channels to server-side compute and storage hosted in public and/or private data centers, and all of it with dependencies on other SaaS offerings. There are unique security challenges facing every layer of the solution. 

Read the full article here

E-Learning for Industry 4.0: Preparing the Workforce for Advanced Manufacturing Automation

By Mark Allinson

In the wake of the Fourth Industrial Revolution, commonly referred to as Industry 4.0, the manufacturing landscape is undergoing a profound transformation. 

Read the full article here
Industree 4.0 is exclusively sponsored by SAP