From AI Tool to Production System—Two Perspectives on the Future of Artificial Intelligence
Artificial intelligence has made its way into many companies—though often still in a limited form. Texts are generated automatically, and initial analyses are supported. But the potential goes much further. This is demonstrated by the assessments of Prof. Dr. Jürgen Seitz of Hochschule Offenburg and Dr. Felix Kalkum, a business professional at BurdaSolutions.
Prof. Dr. Jürgen Seitz: AI Must Be Integrated into Processes
For Prof. Seitz, the biggest challenge in the coming years will be to bridge the gap between the capabilities of modern AI systems and their actual use in society.
“In many companies, AI is primarily used for text generation,” he explains. “Yet models have long been capable of supporting decision-making and taking on entire tasks—provided they are meaningfully integrated into processes.”
So-called AI workflows play a key role here. By combining different models and connecting them to corporate data, agent-based systems are created that can handle entire task chains—from analysis to implementation. This reduces the workload on employees, boosts productivity, and enhances competitiveness.
At the same time, Seitz emphasizes the importance of AI sovereignty for Europe. Instead of relying on individual international providers, he says, an intelligent mix is needed: using international models, but combining them with European infrastructure, European standards, and regional hosting.
“Sovereignty does not mean isolation,” says Seitz. “Promising approaches are already emerging from Europe’s research and startup landscape.”
In the long term, he sees the greatest impact not in a single technology, but in the interplay of AI assistants, autonomous agents, and robotics.
He also expects major changes in areas such as media, education, and medicine—for example, through methods for identifying authentic content, curated learning, or on-device AI for greater stability and data security. Positive real-world examples that build trust are crucial, he says.
Dr. Felix Kalkum: Applications Are Becoming More Important Than Large Models
From a business perspective, Felix Kalkum sees the greatest impact not in ever-larger models or data sets, but in concrete applications.
“It is not models or data in and of themselves that will shape economic development, but rather solutions that solve real-world problems,” he emphasizes.
This focus will lead to open-source models and smaller AI systems gaining in importance. They are more cost-effective, more resource-efficient, and perfectly adequate for many use cases—often more relevant than the latest developments in large frontier models.
Looking ahead to the next five years, Kalkum also expects new model architectures that require significantly less human-generated training data.
“Models trained with less than one gigabyte of human data could already achieve human or even superhuman performance in many tasks,” he estimates.
Particularly exciting, he notes, is the decoupling of AI performance from human input—for example, through self-learning, synthetic data, or interactions with the environment. This shifts the central question:
No longer: “What data do we have?” But rather: “What goals, framework conditions, and values are we optimizing?”
The design of feedback systems, control mechanisms, and reward functions will thus become a key competency for businesses and society.
The public debate will also change: moving away from pure data protection toward the question of who defines the—often implicit—norms by which autonomous AI systems should operate.
Two Perspectives—One Common Core
Whether from a University or industry, both perspectives make it clear that the future of AI lies less in spectacular individual technologies and more in their intelligent application, integration, and design.
Where science and practice collaborate, solutions emerge that are not only technically feasible but also create real value.
Contact
Prof. Dr. Jürgen Seitz,
Hochschule Offenburg
juergen.seitz@hs-offenburg.de
Dr. Felix Kalkum,
BurdaSolutions
felix.kalkum@burda.com
Knowing What Will Be Important Tomorrow
Sign up to receive insights into our research and forward-looking solutions for industry, business, and society.