More than just hype?

Demystifying AI in Customer Support: Understanding the Potential and Limitations of the Technology

The thought of artificial intelligence evokes mixed feelings in many people - between the hope of a smart assistant and the fear of not being able to keep up with the new technology. On closer inspection, the apparent mystery of AI turns out to be a welcome opportunity for the future, which still needs time to mature in many use cases.

Image for article about AI in Customer Support

What this post is about

  • Role of AI in Customer Service
  • Limitation of AI
  • Data protection-compliant setup of AI
  • Experiment with caution

Artificial intelligence describes computer-based, self-learning software systems that mimic strategic thinking and language skills and continuously and largely autonomously evolve. This gives users the ability to solve simple tasks more quickly and save time.

However, the ongoing hype around AI also stirs fears and concerns: Will AI make human labor obsolete in a fully digitalized future? A study by the Boston Consulting Group (BCG), however, proves that this future is still far off. The findings from the study put the hype around AI into perspective and highlight the need to carefully assess its use depending on the application.

AI has arrived in customer service

A year ago, we already took a closer look at the topic of AI in support – a lot has happened since then. Artificial intelligence is already being used in customer service in many areas today. Some of the best-known forms include chatbots that automatically answer common questions and FAQs. Here, there are benefits from the faster processing of recurring inquiries, datasets are searched more quickly, and with each processing of customer data, the system learns independently.

Furthermore, the use of AI in ticket systems and CRMs is a time-saving factor. Pre-set assignments are automatically evaluated, allowing for higher ticket processing in a short amount of time. Does it sound like the use of AI only has positive aspects? Don't let the euphoria fool you.

When things get complex, limitations emerge

When AI is used to solve routine tasks, the technology proves to be a valuable aid. Chatbots or answering recurring questions are tasks that artificial intelligence can handle effortlessly with the right preparation. This is particularly sensitive when it comes to sensitive data that allows conclusions to be drawn about the identity of individuals, as the example of Italy shows: The data protection authority there repeatedly confronts ChatGPT with data protection concerns and threatens to block the service.

According to the BCG study, 90% of participants showed success in creative tasks when they used AI for support. However, the solution of complex tasks that require individual demands pushed AI to its limits. 23% of the study participants performed worse or delivered incorrect results when solving their tasks with the help of AI.

Human before Machine: When it comes to detailed and individual, not universally applicable tasks, one's own ideas and thoughts have more value than digitally generated output. Compassion, critical thinking, ethics, and other skills that customer support teams bring to their work cannot be replicated by AI.

Privacy-compliant setup of AI

Data protection is one of the central challenges in dealing with artificial intelligence. The ability of AI systems to collect, analyze, and learn from large amounts of data raises significant questions about privacy and the security of personal information. This is especially critical when it involves sensitive data that can reveal the identities of individuals.

Legal regulations such as the European General Data Protection Regulation (GDPR) set strict standards for the processing of personal data by AI applications, including the need for clear user consent and transparency about the algorithms used. It also guarantees the right to erasure and rectification. In practice, however, this is difficult as information from LLMs cannot be deleted. Instead, subsequent rules must be introduced and AI applications must be retrained so that they no longer use certain information. Companies must therefore be able to fine-tune their AI applications themselves, not just the providers of the AI models.

Although the EU was the first in the world to create a comprehensive set of rules for artificial intelligence with the AI Act, the topic of data protection and the areas of application of AI remain under discussion. The new regulations will not come into force until spring 2026 and could potentially be outdated in two years' time.

Companies and developers are thus faced with the task of designing AI technologies in a way that is not only effective but also compliant with data protection laws.

Experiment with caution

Personalized customer service is considered the gold standard today, and AI presents itself as a valuable support. However, it also poses a challenge as it needs to ensure data protection while enabling personalized experiences. This requires a balanced and thoughtful approach.

Therefore, Zammad approaches the complex field of AI with caution and well-considered steps. Instead of integrating AI into processes recklessly, we focus on carefully secured, small milestones. We are aware of the technology's potentials, yet it is essential to define clear boundaries within which we can operate safely and creatively. By adopting this approach, Zammad aims to combine innovation with responsibility and elevate customer service to the next level, without compromising on data protection.

Summary

As fascinating as the world of artificial intelligence is, we are only at the beginning of its development. We must not lose sight of this fact, despite all the enthusiasm, from both a technical and legal perspective. Blindly integrating artificial intelligence into projects aimed solely at positioning at the forefront poses too many dangers. It is more important to carefully identify and define specific use cases that genuinely provide added value and support organizational goals. The preparation and cleaning of data needed for AI deployments are also crucial, as the quality of the input data significantly determines the quality of the results.

  1. AI has arrived in customer service
  2. When things get complex, limitations emerge
  3. Privacy-compliant setup of AI
  4. Experiment with caution
  5. Summary
Signup
Together we turn your customers into fans.
Start free trial!
Newsletter
All releases and news directly in your inbox.
Subscribe to the newsletter