AI won't be saving money or employees any time soon
Industry is also hoping that generative artificial intelligence (AI) will lead to great leaps forward. The Chief Digital Officer of Körber AG, Christian Schlögel, is one of the proponents. In an interview with ntv.de, he explains what needs to be considered when using AI in the old economy, where Germany stands and why he believes the technology is indispensable. The questions to Schlögel were completed with the help of ChatGPT.
The international technology group Körber has its roots in mechanical and plant engineering. In the four Business Areas "Digital, Pharma, Supply Chain and Technologies", the Hamburg-based company now also offers software and other digital products with a focus on artificial intelligence. Körber AG is wholly owned by the Körber Foundation.
ntv.de: What do you think of the Altman chaos with OpenAI and the fundamental question of whether AI should be a common good for the benefit of humanity?
Christian Schlögel: I do believe that AI should be a common good. It is such a groundbreaking technology that will change so much that we would do well to make large parts of it generally accessible. So I think the basic idea of OpenAI is right, but it's not sustainable for a company in the long term because it requires massive investment. This also explains much of what has become known about OpenAI in recent days.
You also expect generative artificial intelligence to cause major upheavals in industry. What can it achieve there?
AI has long been used in industry in the form of machine learning to optimize tasks with the help of data. Now generative AI is being added, which creates new things, i.e. primarily supports mental work.
For example?
Let me give you two examples from our company: In paper processing, artificial intelligence can generate recommendations from a large amount of data from the production lines as to how the machines need to be set in order to increase output, for example. Paper is a natural material with many variations in texture. There are more than 300 machine parameters to optimize the result. Machine learning is particularly helpful here. Instead of increasing output, energy consumption can also be reduced. We use another AI application in the pharmaceutical industry: Each vaccine vial has to be checked individually. Here, image recognition can reduce the rate of incorrectly automatically rejected vials by 90 percent. This is because the AI recognizes, for example, that it is only a scratch on the glass and not a foreign body in the vaccine.
Do you have another example of generative AI? These are primarily machine learning.
The use of generative AI is still at a very early stage. However, companies are already using it to write internal audit reports, which they call AuditGPT, based on ChatGPT. The system receives the audit context, the intended structure and the audit result and generates the audit report from this. All this takes just a few seconds instead of many hours of work. Another use case comes from customer service: a customer reports problems or has questions about technical systems. The generative AI, which has been trained with service tickets processed in the past, product documentation and existing knowledge databases, generates suitable solution proposals for even complex problems in just a few seconds.
In her opinion, AI in industry could also make a significant contribution to climate protection. But what about the power consumption and CO2 emissions caused by the AI itself, simply due to the computing power required? How much climate protection is there at the end of the day?
It is important that data centers in which AI is calculated are operated with green electricity. And: the benefits must be greater than the costs. In order to achieve this, massive research is being carried out into how the necessary data volumes can be reduced, to the point where AI models may even be able to be calculated on a smartphone in the future. In the pharmaceutical sector, we train models in the cloud, but then bring them down to the machine. This is necessary for regulatory reasons, but also saves energy.
How and to what extent is generative AI already being used in industrial production here?
Still very little. There is the phenomenon that generative AI is still hallucinating, i.e. outputting things that are not based on facts. We still have some way to go. Some companies such as Mercedes-Benz and others are already using the first prototypes with generative AI, but they are still at an early stage.
How are these companies using generativeAI specifically?
Mercedes uses AI in the area of quality management. The system has been trained with quality data from production, development and customer experience. This allows errors to be identified and rectified more quickly. Another advantage of this approach is that communication can take place in dialog and in natural language without the need for specialist knowledge or programming skills. The system can also prepare the data according to the specifications of the person involved, which can lead to faster troubleshooting. In this way, a kind of conversation takes place between the engineer or service employee and the system in order to understand the problem better and in a more targeted manner.
Where does the software behind it come from, the USA?
Yes, above all. Aleph Alpha from Heidelberg also plays a role in Europe, as local companies pay more attention to data protection, for example. But US companies such as Microsoft naturally also meet these requirements. When AI extensions for Outlook, PowerPoint or Teams, for example, are launched on the market, user data must not be leaked onto the internet.
Which countries are leading the way in the use of generative AI in industry?
USA, USA, USA, then at some point China and then Germany.
Can Germany catch up?
Of course, and Germany must seize this opportunity. Aleph Alpha's recent financing round of 500 million dollars gives us hope, but is still no match for America. What is needed is an effort between universities, start-ups and industry. Innovative companies need fields of application in industry, but also in public administration, where automation has extremely high potential. That's why I'm optimistic that Baden-Württemberg is using an AI administrative assistant from Aleph Alpha to help with research and the creation of texts.
Where is the application of generative AI still lacking in industry?
There is still a lack of understanding of the technology. Americans and Chinese are more open to technology, they try it out and if it doesn't work, they use their findings for something else. In Germany, the goal is often the 100 percent solution right from the start. There is also a lack of venture capital in this country to make companies big. Elon Musk, on the other hand, is investing billions in his AI company.
What is the situation outside Germany?
Generative AI is still a very young technology and has achieved a breakthrough in a very short space of time with ChatGPT. This is because a massive hurdle has been removed: anyone can now simply talk to the system without any programming knowledge. This also came as a surprise to large corporations such as Google, who wanted to perfect their own system before it was launched on the market. But suddenly everyone had to follow suit. For the software giants, this is a must because it will massively change all of their offerings: We will see AI in the form of co-pilots for all applications.
What regulations apply to the use of generative AI in this country - under what conditions and how is the industry currently allowed to use it?
The co-determination rights apply as with any technology, as does the General Data Protection Regulation. The EU AI Act is currently being formulated, as is the EU Data Act, according to which customers can decide what happens to their data that is generated by machines. I don't see any major legal hurdles. The questions are more about how to train AI, how to obtain data and how to involve internal company groups such as the works council. The EU will also introduce ethical rules to ensure that data used to train AI systems does not contain any prejudices, for example. However, I see this problem less with technical data for machine control. Nevertheless, companies like our customers take it very seriously, which is also important for employee trust.
Do you still see dangers in the use of AI in industry?
I see more opportunities than risks. The biggest risk is that people are too eager to fully automate business processes with the help of AI. You need several loops to ensure that the AI produces the desired results. In my opinion, Microsoft's concept of a co-pilot that helps me to be more productive is a good one. As a human being, I am still in charge.
What do you say to critics' concerns that AI could dominate us and even wipe us out?
The concern that systems will achieve complete autonomy and then no longer be controllable comes from the fact that we don't yet fully understand why the system reacts in this way or that. If we stick to the mantra of the co-pilot, I don't see this danger, and certainly not in industry.
Are there still ethical concerns when using the system in industry?
Yes, if deepfakes suddenly cause people to say things they never said. The fake email from the supposed boss to quickly make a large bank transfer can turn into a video call from an avatar who looks and speaks like your own boss. This danger also applies to industrial companies. Processes must then be found to protect against it. Wherever personal data is involved, I also have to pay attention to an ethical framework, for example in customer contact.
Which and how many jobs in industry could generative AI replace and within what timeframe?
That's difficult to estimate because it depends on how the technology develops. For Europe, however, with its ageing population and the change in the skilled workforce, it is more of an advantage. After all, we need to increase our productivity in order to secure our prosperity. AI will make employees more productive, especially when it comes to mental work. Fields of work will change as a result, but that has always been the case: some tasks will disappear and new ones will be added. In the automotive industry, over 90 percent of body shell production is now automated. This has not meant that industrial workers are no longer needed, but new, higher-value jobs have been created to operate such systems. We will see the same thing with generative AI. We will automate more routine tasks and be able to use our human capital for more creative tasks.
How many and which people will lose their jobs as a result, and what alternatives are there for those affected?
It's difficult to quantify, but yes, there will be people who lose their jobs. We then have to see how we can get these people into other jobs. That will also be a social task. This includes a great deal of further training - where AI can again help: through personalized training content.
How much can industrial companies reduce their costs through generative AI?
It's too early to say, because we're only at the beginning. First of all, we need to look at which tasks the AI systems are suitable for and where the investment is worthwhile. According to a Forbes survey from the spring, most American companies do not expect any major cost changes or effects on the number of employees in the next two years. Now is the time to find fields of application.
Christina Lohner spoke to Christian Schlögel
The Chief Digital Officer of Körber AG, Christian Schlögel, believes that AI should be a common good for the benefit of humanity, acknowledging its potential to cause major upheavals in industry. Generative AI is being used in the company to create recommendations for optimizing production lines and reducing energy consumption in paper processing, as well as to reduce the rate of incorrectly rejected vaccine vials in the pharmaceutical industry.
As industry continues to hope for significant advances from generative artificial intelligence (AI), concerns about the labor market arise. Christian Schlögel suggests that while AI will replace certain jobs, it will also create new, higher-value tasks, indicating a shift in the workforce as a result of AI integration.
Source: www.ntv.de
