The AI train is rolling

How do you see the current developments in machine learning?

Prof. Peter Lehmann: The topic of machine learning and artificial intelligence (AI) has been around since the 1960s/70s.

But the conditions have never been as good as they are now.

First, we have user interfaces that any user can operate. So I don't need to be a programmer to do machine learning.

Secondly, the technical infrastructure that scales is available, algorithms come from the cloud and the licenses and machines are also ready. So it has never been so easy to use cloud computing or software-as-a-service.

Thirdly: I have the pressure to take action on the subject of prediction because my competitor is doing it, especially if I am an international company.

Asia, the U.S., and the U.K. in particular are much further along in this regard. At the latest when my competitor makes progress on the basis of predictive analytics and adapts better to its customers, then I will be forced to take action.

What then prevents German companies from taking action?

Lehmann: The necessary skills are missing. So there is a lack of people who can actually set up the infrastructure.

But I think it will really take off in the next three years. The major manufacturers are all already on their way.

At what level are decisions made based on artificial intelligence?

Lehmann: I think this will be especially prevalent in new business development, or adapting and developing new products, to finding and penetrating selected markets.

Do we still need gut feeling or human intuition?

Lehmann: It takes gut feeling to verify the results found - so that still has to be done by humans. It is quite amazing how many decisions are made based on gut feeling.

The difficult thing is to make so-called fuzzy decisions. US President Obama once said: I only have shitty options, and the worst thing is to do nothing.

So the decision to choose the best among bad choices, that's very difficult - and this is where gut feeling is still needed.

What challenges do companies currently still face when they want to implement Big Data projects?

Lehmann: Big Data is not primarily about processing a lot of data. Traditional and established databases can do that, too.

Big Data is about making data with different formats, such as images, videos, texts, audio, analyzable and thus usable.

For example: Is the customer who has just vocally struggled through the call center menu in a good or bad mood?

To retain the customer, it might now be better to use a male or female and trained voice to take the call. So, when you're dealing with Big Data, you also have to be clear about what benefit you're trying to achieve.

What role does SAP play in AI and machine learning with Hana?

Lehmann: A big role. So I think SAP will penetrate the existing base. That's their business, that's what they live on.

SAP is very powerful and has done a lot. In recent years, SAP has developed into an American company. That means that a lot of the know-how that is available in Palo Alto is now also available in the software, and SAP is very strong in the prediction environment.

It remains to be seen what will happen with BW. BW for Hana will continue to play a major role for some years to come. Very many prediction providers support SAP Hana or even BW accesses.

And we see BW increasingly migrating with the new objects in Hana.