The rise of decision intelligence: AI that optimizes decision-making
Today, doing more with less is a key principle that drives business strategy across many resource-intensive industries. Businesses are looking to get a higher return out of artificial intelligence (AI) and machine learning (ML) than just great insights. They need access to recommendations that help simplify complex decisions around how scarce resources should be allocated, how to schedule tasks, and how to deal with constraints. As Alex Fleischer points out in his blog, a recent Enterprise Strategy Group (ESG) technical validation report cites the need to improve operational efficiency as the overarching theme driving AI and ML interest.
To learn more, I sat down with Virginie Grandhaye, offering manager for IBM Decision Optimization. Virginie has been bringing products to market to help our clients optimize decisions at speed by combining human and machine intelligence.
Virginie, have you always been interested in analytics? Can you tell us about how you first embarked on your journey?
I’ve been working in the advanced analytics domain for 15 years. I have a master’s degree in applied mathematics and worked on analytics software quality assurance most of my career. In my current position as offering manager for several products within the IBM Decision Optimization portfolio, I focus on providing quality software products to help drive customer success. I also meet with many business partners, clients and prospects to understand their business, gather feedback on products, and evangelize about the benefits of operations research capabilities applied to the business. This allows me to bring to bear my technical and business experience.
What excites you the most about your role as an offering manager bringing AI products to market?
I have always been fascinated by the power of mathematics to solve industry problems. When I travel to conferences, I hear clients describe the complexity of their typical decision process and their constraints, and how it takes days or weeks for them to find a solution.
Many clients tell me that with IBM CPLEX, they can find a solution to their problem in much less time, thereby driving huge ROI. Hearing satisfied clients talk about the value of optimization excites me the most.
When you look at the key requisites for success in AI, how do IBM data and AI offerings uniquely help address client needs along their AI journey?
I’ll start with my own definition of AI as applied to the business of decision-making. I see AI as enabling the use of data to both analyze and formalize the decision-making process and optionally automate the decision. Based on this, I think there are several ingredients needed for a successful AI journey:
- A good understanding of the business problem you want to solve.
- Deep knowledge of the various techniques, such as machine learning and optimization, and an understanding of how to combine them to solve your business problem.
- Alignment with the business team—and this is critical. Success happens only when working by iterations, with the decision-makers involved. To transform the decision process, alignment is mandatory.
- A strong data science platform to handle challenges such as data governance, accuracy of models, validation of assumptions and visualization of data.
- The capabilities to successfully deploy. Again, a good platform that simplifies deployment is a plus. It should include the ability to monitor and manage the model life cycle. There is an evolution over time. The accuracy of models change when they encounter production data, and that needs to be taken into account from the beginning.
Increasingly, we are seeing the spotlight on decision intelligence, or the use of models for applied decision-making as part of AI. What do you advise businesses who are seeking to use AI for decision support?
I like the fact that we now have a term for AI applied to the business: decision intelligence. Indeed, a decision intelligence framework helps with operationalization of AI or ML for real business decisions. And I think it was needed.
We see a lot of myths about what AI can do, but not enough success stories on how people infused AI in their business. In my position, I have the opportunity to meet many smart leaders, who have invested in advanced analytics with IBM. These are business leaders who have a need for a business concept, not a mathematical one. And that is what decision intelligence brings to the table: it is a focus on the business need instead of the types of algorithms that machine learning relies on.
Now, to answer your question, I strongly believe that decision intelligence is something that every business leader should know about. That is why I’m also invested in teaching about AI, ML, and digital transformation – applied to the business, in high schools, both for engineers, and business schools – here in France.
We cannot make every CEO an expert in classification methods or linear programming, but we can make them aware of how data and AI can help them achieve in arriving at optimal decisions.
Tell me about client success stories that used decision optimization and artificial intelligence.
Depending on the industry, we see tactical, strategic, or operational use cases coming into play. Many clients in industries such as transportation, supply chain and financial services have benefitted from huge cost savings, as well as benefits such as minimized delivery times and maximized production volume. Optimization solvers like IBM CPLEX take complex decisions and solve them in a matter of minutes or hours compared to days to weeks. And they accommodate last minute changes. Decision optimization software can help avoid costly rework when business dynamics change.
One customer was able to reduce annual bulk transportation mileage by more than a million miles, saving millions of dollars per year.
There are simple use cases that are easy to start with. For example, portfolio optimization can be applied in financial services, food, and even the energy sector. Price optimization serves airline, hotel, and pharmaceutical companies, among others. Predictive maintenance applies to manufacturing companies as well as to banks who have ATM maintenance.
That is another reason why I find this domain so interesting: The same mathematical strategy can be applied to solve business problems for many different industries.
Where do you see your decision optimization offerings going?
To create a strategic plan or a production schedule, clients need access to predictive insights that can come from machine learning models, and prescriptive insights to guide them on their best course of action.
IBM Watson Studio is the unified environment that simplifies the development of machine learning and optimization models. Once it is in place, both predictive and prescriptive models can be deployed using Watson Machine Learning.
All the offerings mentioned above are available in IBM Cloud Pak for Data. I strongly believe that this is the way people will modernize their IT. They will benefit from all the core data microservices, and the ability to scale up or scale down in a containerized architecture, using the cloud without having to move their data. Thanks to IBM’s acquisition of RedHat, I believe that we have a competitive advantage, especially with our multi-cloud capability, provided by Cloud Pak for Data. We are expanding our investment by developing a great experience for our users.
In addition, we are making optimization modelling activities easier. For instance, our award-winning AutoAI capability uses AI to build AI, helping people achieve value from machine learning faster. And we recently developed a modelling assistant for optimization users, to simplify the model building process using natural language inputs.