Artificial intelligence to boost Smart Manufacturing Myths and realities
Dr. Athanasios (Thanos) KONTOPOULOS, Computational & Data Science Scientific Director and Global Lab Leader, Air Liquide [EPA: AI]
Dr. Athanasios (Thanos) KONTOPOULOS, Computational & Data Science Scientific Director and Global Lab Leader, Air Liquide [EPA: AI]
There are many discussions and, let us say it, much buzz about this topic. People in Industry, even more in Manufacturing, are requesting realities beyond the myth. Especially if and how they can boost their business with this technology.
Where can we apply it?The applications of AI are numerous. In Air Liquide, we use this framework for our digital transformation strategy: Assets-Customers- Ecosystems (ACE).
For the Assets: predictive maintenance, process, and energy optimization, energy management, energy storage, logistics, supply chain, plant design are some of the fields where AI can push the business forward.Finally, machines are not emotionally affected when it comes to rational decisions like the ones related, for instance, to stock trading
However, let us beware of the myth that wants to throw away the “old” technologies such as operations research or computational science and replace it by deep learning. Instead, we combine the best of both worlds. It is feasible, we have already obtained results and we are part of a cross-enterprises collaborative project that will push this “hybridization” frontier forward.
For the Customers: recommender systems, customer analytics, pricing, churn, demand forecasting, building new types of contracts... are just some of the applications where AI can incredibly help us to make better decisions or automate processes and improve the Customer experience. We have obtained very positive results with AI in most of those fields.
Nonetheless: can everything be forecasted thanks to AI? Our experience shows that we can have successes, for instance, in demand forecasting, but some phenomena are either too complex or we lack information (data) to be able to predict them. So expectations need to be set at the right level, according to the case.
Let us now deal with Ecosystems (how employees interact internally/ externally).
Humans and machines - some more myths and realities Let us now see the strengths of humans and machines-AI in various fields, also beyond manufacturing.
If you seek purpose, when you need to put things into context and use common sense, do not go and ask an AI system. E.g., when you’re building a strategy, AI can help you to some extent but no more. It is your call. When working on innovation and needing creativity, AI can be of service (e.g., in the discovery of drugs, new materials...) but do not expect the next “big thing” in your business to happen by pushing on an AI button. You’re in the driver’s seat.
Thanks to the ability to put things into context and have common sense, as we mentioned before, human can deal more efficiently with reaction to significant change. Again, machines- AI can help but in no way, they can take alone these major decisions, e.g., in crisis management. Nevertheless, when we are talking about processing big data information in a short period of time (usually almost instantly), machines can deal quickly with this. Think about, let’s say, the recommender systems, a huge volume of processed data for a simple effect perceived by users! Finally, machines are not emotionally affected when it comes to rational decisions like the ones related, for instance, to stock trading.
A myth: we do not need people any more. I believe the reality is that we can “augment” people with AI and allow them to save time / be focused on added value tasks. Humans and machines- AI can complement each other to make incredible achievements, while putting the human at the center!
Another myth: AI Tech is enough to succeed - you just need to have the best “algorithms” in the market! This is one of the most common (and most dangerous) myths.
If you want an Organization to succeed with AI, the first that is needed is to stay close to the business needs. Get the business cases right. This does not mean to forget tech: seeking scientific excellence, testing, and using cutting edge technology is a must. However, do not go out using a complete techno-push approach.
Once you have your business cases, then you might want to use a “3-D” approach:
● Make sure to have the adequate skills (people). Technical skills (in data science, data engineering, IT architecture and programming, design ) are essential here but soft skills are equally very important. think about training, communication, connection business...
● Governance and Strategy: the Organization needs to evolve. Functions such as data owners, data officers and central data governance bodies are important. AI needs data to be based upon.
● Infrastructure. Data lakes, namely, to store data in a structured manner and access them easily and cloud technology. A clear must.
The most usual myth here is to think that success is reached by having just one or two of the elements above. It could never work. You need them all.
Conclusion
Humans and machines-AI can work incredibly well together. Success comes with more than AI technology. Let us think of AI as “Augmented Intelligence” and put the human at the center.
Weekly Brief
Read Also
