Machines aren’t taking over the world—not quite yet—but businesses that harness machine learning technology certainly have an advantage. Machine learning is globally recognized as a key driver of digital transformation, which according to SAP, is about finding new ways to deliver value, generate revenue, and improve efficiency. Intelligently, of course.
The SAP Leonardo Machine Learning Foundation was introduced two years ago to help “enhance business processes with intelligent algorithms using enterprise-grade machine learning capabilities.” The foundation allows users to build an intelligent enterprise through applications—both functional services and core capabilities.
We sat down with Sebastian Wieczorek, vice president of SAP Leonardo Machine Learning Foundation to ask how it works and how ASUG members can benefit.
Sharon: The SAP Leonardo Machine Learning Foundation was launched nearly two years ago at SAPPHIRE NOW and ASUG Annual Conference. Can you give us a brief overview of what it is and how it has evolved?
Sebastian: We started out with package services—called functional services—so that people could begin using machine learning without having data science experience. They could also use our APIs off the shelf and vet them.
What we have done since then is evolve the platform by opening up more of the functionality that we are using to develop these services. Now you can build your own machine learning models and deploy them on the foundation. You can use them as you use the services that we’re providing, but you can also now deploy your own training and build your services, opening up the whole chain of infrastructure and tooling that SAP uses to provide its intelligent enterprise functionality.
What we announced at SAPPHIRE NOW and ASUG Annual Conference is the next evolution of SAP Leonardo Machine Learning Foundation called SAP Data Intelligence. We merged SAP Leonardo Machine Learning Foundation with SAP Data Hub as a cloud solution so that customers can leverage the combination of data orchestration and machine learning functionality.
Sharon: Can you give us an overview of SAP Data Intelligence? What is it aiming to accomplish?
Sebastian: SAP Data Intelligence came out of a joint project between the SAP Data Hub team and the SAP Leonardo Machine Learning Foundation team with support from the SAP Analytics Cloud and SAP HANA teams. The intent was to create tooling on top of the isolated assets that we have in the two platforms and provide one integrated solution for data scientists, developers, and operations teams to build the machine learning models and handle life cycle management on top of that.
Sharon: How can the SAP Data Intelligence help customers who want to apply machine learning within their organizations?
Sebastian: I’ll give an example of how one of our customers has used it. Daimler, a German multinational automotive corporation, was looking to accelerate the conversion of its car sales. We co-innovated an intelligent app that enabled car model identification through a photo. A customer takes a picture of a car they want and then machine learning identifies the car model, the color of the car, the rims, and so on and then finds a nearby dealership that can facilitate that purchase. The outcome was an improved customer experience.
Another project that we worked on with Daimler was taking sales data out of SAP HANA to predict the value of used cars. In both these use cases, we were able to implement data intelligence on top of the machine learning capabilities that SAP provides in SAP HANA, in SAP Leonardo Machine Learning, and in other places.
SAP Data Intelligence is the one-stop shop for developing one project after the other for your digital transformation and machine-learning enabled company strategies. It gives organizations the tool sets for building, for defining end-to-end processes, and for connecting all their data sources, whether they are SAP or non-SAP data sources, and whether they are structured or unstructured.
When you look at it from a strategic perspective, the CIO of an organization is concerned with who has access to the data, with pre-processing, with being able to explain a model’s result, or trace back to a particular version of the model and its training script. SAP Data Intelligence is one platform that allows them to manage all of that.
Sharon: What are some success stories you can share with us about those implementing ML via the SAP Leonardo Industry Innovation Kits?
Sebastian: The SAP Leonardo Innovation Kits are basically vehicles for making it easier for customers to leverage SAP Leonardo capabilities, and machine learning was one of the central parts of many of these innovation kits.
When an organization wants to know how to start its SAP Leonardo adventure, it’s very helpful to have SAP provide the blueprints, or projects that they could try out. But, of course, what we see right now more and more in customer conversations is that customers already have a strategy for machine learning in place. Unlike two or three years ago, they now have their own machine learning teams, they know what they want to do, and they have complete cases where they know where the business value lies.
So, for them it’s not just the looking into cases that SAP is blueprinting, but they have their own cases and for them these kits are interesting to get additional ideas. But underneath, they still need a platform and an innovation system to drive their own cases as well. That’s how these two parts relate.
Sharon: Have you come across situations where you’ve added a customers’ use case to the SAP Leonardo Innovation Kits?
Sebastian: That’s a good question. We see that quite a bit. From a development perspective, we support customers directly and look into certain use cases when we feel that there’s potential to scale that out to more customers.
When we look at the Daimler case, for example, recognizing your own products in a picture is a need that is very common. We have a jewelry producer that is now rolling that out internally as an app for its maintenance and support center to detect what is the product that customers are sending in that is broken to find what is the material it is made from and who is the expert at fixing it. In a maintenance case where machinery is broken, workers are taking pictures to order the repair parts that they need. The same solution can help drive conversions in the sales process. There are a lot of different use cases behind this technology, but the basis underneath them all is a simple machine learning case.
Sharon: You’ve spoken before government bodies about the ethics of artificial intelligence. Are you concerned about the future? What, in your opinion, can we do now to prevent the robot apocalypse?
Sebastian: I’m an appointed expert for the German parliament and I also serve on the ethical board for artificial intelligence (AI) for SAP. It’s a topic that’s interesting for me personally, because I want to work for a company that is taking this seriously. Many believe that when a corporation talks about AI ethics, it’s simply because that corporation wants to justify its use in front of the public. It’s so that they understand that we’re not doing evil. And it’s, of course, a valid case for a company to make its position clear that is not the intention.
But for me personally, I feel it’s important that our own people (internally) know that the company has values. It shouldn’t be just one person charged with deciding what is morally OK and what isn’t. And so, I think it’s very important that societies are defining what is acceptable for them, and that companies are explaining what is acceptable for them, and where the red lines are.
Now, you asked whether I’m concerned. I believe that there is a lot of hype. Often, people don’t understand what the technology is, what it is capable of, or what it actually isn’t capable of.
I think that AI ethics will be a part of business ethics. Ethics in general is a philosophical topic, right? You have a lot of science over the history of humankind that has come up with a lot of different frameworks of how to deal with ethical issues, and I think that AI ethics is a special case in terms of technology allowing us to do things that were not possible before. But if we all understand what’s now possible, we can map that back into these frameworks. This is what we’re doing right now. This is, of course, work we are conducting together with people who have a degree in philosophy, and people who have a degree in business, and people who are on the engineering and on the data science side.
But I have the feeling that when you go into this open discussion and you also take the effort to explain to people what it is, what it can do, and how SAP as a company along with we as a society can and should use this kind of technology, then we’re also leaving some of these fears that are attached to the discussion aside and really focusing on what kind of behavior we want.
Sharon: What’s the one thing you would want ASUG members to take away about the short-term future of SAP Data Intelligence? Why should they care now?
Sebastian: I think that data intelligence is the next logical step for us as a company and for our customers. When you think about data science and AI in general, and the effort it takes to implement a project, it’s probably 20% building the algorithms and getting them to work, and 80% to secure the process of data access. It’s mostly securing the process of life cycle management and operationalizing the model that you are creating so it runs in a production scalable manner when you put it in the hands of your customers or your internal workforce, or you’re running your mission critical processes with it.
Data intelligence is not just a vehicle for you to build one or to solve one data science problem. I see data intelligence as a strategic decision for a company to create a system on which it is innovating with data science and non-data science projects whenever it wants to draw value out of that data.
Sharon: What’s next for SAP Leonardo and machine learning?
Sebastian: There are a lot of nexts. We’re introducing SAP Intelligent Robotic Process Automation to the market. We’re working very hard at bringing more and more intelligence into all the solutions that we have. I think we’re also going to see that more and more customers are now taking up the technology platform that we are building and creating their own use cases. And, for me personally, I am happy to see that the technology that we are introducing to the market is getting the attention it is. The exciting part is to now see its adoption, especially in products like SAP S/4HANA, SAP C/4HANA, and so on. I’m happy to see that we’ve reached the point where it’s becoming mainstream now.
Interested in learning more about machine learning and what it means for your data strategy? Register for “How SAP Leonardo Machine Learning Supports Finance Cash Application,” webcast.