The introduction to Artificial Intelligence your business cannot miss
The History of AI
Alan Turing is said to be the godfather of AI. He produced the idea of intelligent machines back in the fifties and designed the chess computer Turochamp. However, computers were not strong enough back then, not yet able to make connections or apply smart techniques. It took until 1997 to beat the famous chess player Garry Kasparov. In 2016, the computer AlphaGo defeat a Go world champion. This was a milestone for AI, as Go is known as the most challenging classical game for artificial intelligence due to its complexity.
GTM Director - R&D
GTM Director - Artificial Intelligence
Exploring AI Systems
The purpose of AI has always been to have computers think and react as if they were humans. In other words, it is the creation of software to mimic humans’ behavior and capabilities. The technology evolved over the years resulting in different AI systems ranging from Machine Learning to Generative AI (see graphic below).
In the first case, you will train the system by feeding it data and teaching it how to handle and interpret it. This creates models that will help to interpret new input. The system will do this by comparing the new input with what it has learned through the model. Recognizing invoices and extracting the necessary data from incoming invoices is a well-known use case.
Deep learning, a form of machine learning, creates neural networks that enable the system to learn and make decisions on its own. It requires much more data than classic machine learning, but also a lot more computing power. A significant difference is that classic Machine Learning requires human intervention to help the system make final decisions in case of inaccuracies. Deep learning requires no human intervention to make this decision.
Generative AI goes a step further and does not only interpret data and/or make decisions. Its primary function is to generate content. An example could be a draft of the next presentation you need to create.
Generative AI systems rely on Large Language Models (LLM), designed for understanding and generating content. It will boost workers' performance and is the “Co-Pilot” of every employee. It can generate new texts, images or even a complete Power BI dashboard, using the data you provide as input. The outcome is just as good as the input is. You need to make sure to specifically ask what you need. Garbage in is garbage out. We will dive deeper into writing good prompts in a later article.
Selecting the right approach (machine learning may be sufficient) and model for your task is crucial. Some models are more focused on text, while others are more focused on images. Thereby each AI solution has its own strengths and weaknesses depending on the purpose you want to use it for.
Harnessing Data’s Potential
Training AI algorithms requires enormous amounts of high-quality data. An IOT (Internet of Things) platform such as PTC ThingWorx, PTC’s platform focusing on the Internet of Things (IoT), can be a source of data. But ERP (Enterprise Resource Planning) systems with a lot of historical data can be a reliable source as well. It all depends on the use case you want to optimize/automate with AI.
Imagine your devices break down at random times and the data is captivated by ThingWorx. In this case, AI could search for the reasons and find trends to predict potential future breakdowns. Your data does not necessarily need to be stored in the cloud, but the capabilities of online computing are much bigger than on-premises.
AI’s Impact Across Different Fields
Fields such as business process optimization or worker assistance do not necessarily require AI, but they can derive benefits from it.
Take for example Process Mining (9A business analytics tool discussed in our Dutch podcast), where we analyze the processes of an organization based on data. We track what forms a user opened, how they navigated to another form, and how long this takes compared to other users e.g. With AI, it is possible to automate this analysis and even suggest optimizations.
It becomes even more interesting if we look at Worker Assistance. AI software can e.g., provide field technicians with the required information to complete repairs. With AI, this interaction becomes more human. The technician could just ask: “How should I change the oil?” where Copilot provides step-by-step instructions retrieving this from a combination of various sources available in your organization.
What is Microsoft’s Approach to AI?
Let us start with another example. You are a developer who wants to develop an app. You start from nothing by drafting a basic layout and adding logic to your form. How long does that take you? Image being able to ask AI: “Create an app to add my expenses. I need to be able to add the amount spent, link it to a project, and add a picture of my payment receipt.” In response to this prompt, Copilot in Power Platform draws a basic form with the necessary fields. It might not be perfect, but that is where you come in. The dull work has already been done and now you can tweak it to make it top notch!
That is precisely what Microsoft aims to do - empower you. Microsoft Dynamics 365 Copilot gives you a head start with its AI-powered digital assistant. It allows you to work more efficiently, be creative, and focus on true added value.
Drawing parallels to the revolutionary impact of the steam engine during the Industrial Revolution, this innovation symbolizes the subsequent powerful advance towards heightened productivity.
Note that not all Copilot solutions are available in all regions.
Works alongside you in the apps you use every day
Turbocharge your workforce with a copilot for every job role
Imagine it, describe it, and Power Platform builds it
Defend at machine speed with Microsoft Security Copilot
Increase developer productivity to accelerate innovation
Copilot in Dynamics 365 Field Services
We experimented with incoming emails for customer support. A customer complains about an appliance that broke down.
A shortlist of actions that AI can now take care of:
- Copilot can catch the level of irritation of the customer and assign the proper priority to the email based on the content. It is even smart enough to see through dialects or improper grammar.
- Next step, a work order is automatically created and only needs to be reviewed by the customer support agent.
- The customer also gets confirmation of when the field service technician will come and fix the device.
- The field service technician is extensively briefed on the planned intervention.
Where does Azure AI fit in?
Azure AI is Microsoft’s portfolio of artificial intelligence services designed for developers and data scientists. This way, it is possible to build and deploy your own AI solutions using their models. The speed at which Azure AI gathers information and speeds up line work is dazzling.
Bing Chat (part of Microsoft’s search engine)
Chatbots are best known by the larger public. Bing Chat and ChatGPT both use OpenAI’s language models. However, there is a significant difference as the latter saves chat history to further train its models. That being said, a data leak is never far off.
Choosing Your AI Tools Wisely
You should not dive into AI just because it sounds cool. It is important to evaluate what use case is relevant to you and to prioritize. First, master your processes to see where the added value of AI could be. All tools come with a price. As always, it is important to compare the cost to the potential efficiency gains.
The world of AI holds immense promise for businesses willing to harness its capabilities. From revolutionizing processes to enhancing customer service, the potential applications are vast. By understanding the different AI systems, leveraging data, and choosing the right tools, businesses can unlock the full potential of AI and drive sustainable growth in today's competitive landscape.
Stay tuned for more articles in this series as we delve deeper into the exciting realm of Artificial Intelligence.