Business GPT in public administration: maximising security and efficiency with AI

Public administration faces the challenge of continuously improving efficiency and service quality, whilst at the same time reducing costs and complying with legal requirements. Artificial intelligence (AI) offers numerous opportunities in this regard to automate processes, optimise data analysis and thus process citizens’ enquiries more quickly and accurately. For decision-makers in the public administration of cities, local authorities and regional governments, it is crucial to recognise the potential of AI and utilise it in a targeted manner. However, this also requires addressing the legal, technical and organisational hurdles associated with the implementation of AI technologies. Business GPT is a bespoke generative AI solution developed specifically for organisations in the public sector. Business GPT combines state-of-the-art AI technology with high data protection and security standards. In this blog post, we highlight the key use cases and challenges of generative AI in public administration and demonstrate how our Business GPT solution can help organisations deploy AI securely and sustainably.

• Citizen services: Business GPT can act as a virtual assistant, for example on-site in buildings or as a chatbot via a website, answering citizens’ enquiries round the clock. Enquiries can be answered automatically, thereby increasing the efficiency of customer service without relying on human intervention. Details such as the multilingual capability of language models when providing information to citizens can lead to a massive improvement in service quality and a reduction in the workload for specialist departments.
• Internal administrative support: Staff can use Business GPT to find information quickly and answer simple internal enquiries. This promotes a more efficient working environment and supports administrative tasks.
• Public relations and communication: Business GPT can provide citizens with personalised information via websites and portals, which improves the flow of information and communication and leads to more effective interaction with the public.
• Training and professional development: Business GPT can serve as an interactive learning platform to address employees’ individual learning needs and support their professional development within the administration.
• Improvement management: Business GPT can analyse feedback from members of the public and help to identify suggestions for improvement, as well as develop new services and programmes that meet the public’s needs.
• Human Resources and Recruitment: Business GPT can be used as a screening tool for applicants to speed up and simplify the selection process by initially reviewing applications and answering basic questions.
• Data analysis and reporting: Business GPT can analyse large volumes of data and produce easy-to-understand reports that help to make informed management decisions and improve efficiency.
• Project management: Business GPT can support project teams with administrative tasks by helping to organise tasks, track progress and provide automated updates. This enhances the effectiveness and transparency of project management.

The opportunities offered by the use of Business GPT in public administration are diverse and far-reaching. It enables numerous processes to be automated and optimised. Business GPT’s ability to consolidate unstructured text and quickly provide relevant information leads to significant time and cost savings. This allows staff to focus on value-adding tasks. Public administrations that make targeted use of this technology will be well-equipped for the future. However, there are also numerous challenges and risks associated with the use of generative AI that must be taken into account. 

Legal framework and data protection: Compliance with strict legal requirements such as the General Data Protection Regulation (GDPR) and the Federal Data Protection Act (BDSG) is essential. At the same time, country-specific regulations and the new EU-wide AI Regulation must be taken into account when it comes to the use of AI in public administration. Precautions must be taken, particularly with regard to data protection, liability and the transparency of AI decisions.

As data protection is a crucial factor in the selection of AI solutions, particularly in the public sector, it is important to define the specific use cases in detail and to carry out a data protection impact assessment (DPIA). Telekom has carried out a DPIA for Business GPT, and the relevant information can be provided as part of our service offerings.

For applications that process less sensitive data, providers with more flexible data protection policies may be a good choice. These solutions often offer greater flexibility and can be implemented more quickly. Possible use cases for less restrictive data protection solutions include:

• Data analysis and decision support: In areas where no personal data is processed, less restrictive data protection solutions can be used effectively. Examples include traffic management, the analysis of public infrastructure data, and the prediction of maintenance requirements.
• Publicly accessible services: Applications such as citizen portals providing general information can benefit from the more flexible data protection guidelines. 

For applications that must meet high security and data protection requirements, all statutory data protection measures must be complied with, and robust security mechanisms must be ensured.

Examples of scenarios for highly sensitive applications include:

• Management of sensitive personal data: Applications that handle personal data, such as population registers, health records or social services, require strict data protection measures.
• Internal administrative processes: When automating internal processes involving sensitive data – such as human resources management, financial management or internal communication – high data protection standards are essential to ensure that data is protected.

Specifically for these use cases, Telekom offers a Data Protection Impact Assessment (DPIA) as a key service component, enabling us to comprehensively examine and assess data protection risks in collaboration with our customers. For Business GPT, we also undergo the ‘Telekom PSA process’ to ensure that comprehensive security and high data protection standards are met. Business GPT is based on the Microsoft Azure Cloud, which guarantees that data is stored in Europe. The solution is operated on Telekom’s own infrastructure or on a customer’s own Microsoft Azure tenant. The data used does not leave the secure environment of the application. As soon as Business GPT is no longer in use, all allocated resources are deactivated and the data is deleted.

For bespoke projects operated in customers’ own data centres that require the highest level of control over their own data (protection of intellectual property), Telekom also implements ‘private’ Large Language Model (LLM) solutions ‘on-premises’. Specifically, this involves the use of language models on Nvidia hardware that is operated on a dedicated basis at our customers’ premises. For ‘on-premises’ projects, the focus is particularly on long-term strategic independence from cloud providers (open source), as well as predictable costs (capital and operating costs). 

• Security requirements: However, innovative AI applications also come with significant security requirements. Protection against cyber-attacks is a top priority, as generative AI systems can potentially generate or process sensitive information. Another key aspect is ensuring data integrity. Generative AI must not only be protected against unauthorised access, but must also ensure that the generated content is not manipulated or misused for malicious purposes. This requires comprehensive security measures, such as encrypted data transmission and strict access controls. Regular security audits and updates are necessary to keep the systems up to date and protect them against new threats. Given the rapid developments in the field of cybercrime, security protocols must be continuously adapted.

Telekom’s Business GPT solution is based on a robust operational and support framework. Contractually agreed Service Level Agreements (SLAs) guarantee availability and reliability.

• Transparency and traceability: Public authorities must ensure that decisions made by AI systems are transparent and traceable in order to build trust and comply with legal requirements. With Business GPT, we support the transparent design and deployment of a bespoke AI solution. 
   
• Cultural and organisational barriers: Staff and members of the public should have confidence in new AI technologies. This requires transparent communication and education on how AI works and its benefits. This can be achieved through targeted training and information campaigns. Ethical considerations such as fairness and non-discrimination must be taken into account, and AI systems must operate impartially. Public authorities should develop ethical guidelines for the use of AI and continuously monitor compliance with them.

With the Telekom Business GPT solution, we offer our customers and their staff the best possible start by including training and product demonstration quotas directly within the service packages on offer, with a view to promoting acceptance within the relevant departments and the effective use of business-relevant use cases. Furthermore, we offer a cross-organisational and transparent billing model that is not based on the number of users. This ensures that all relevant and interested employees can benefit from the solution. With a user interface based on a rights and roles concept, business units can be mapped organisationally and managed strategically.

• Scalability and adaptability: The technologies must be scalable and adaptable in order to meet the diverse requirements of public administration. They should integrate seamlessly into existing IT infrastructures and be expandable where necessary to cope with increasing demands. The ability to integrate with existing systems is a crucial factor here, as many public administrations rely on complex and diverse IT landscapes.

Business GPT has been developed to integrate seamlessly into existing IT infrastructures. This simplifies implementation and enables the AI functionalities to be utilised quickly without the need for extensive changes to existing systems. The solution offers an API that enables flexible and scalable integration into existing business processes. This adaptability is a particularly important requirement for public administration, given the complexity of its IT systems.

Addressing these complex challenges is crucial to fully realising the potential of AI in public administration. We support you right from the start as a trusted partner.

In a separate article, we will examine how to establish a suitable data foundation for the effective deployment of AI solutions, and how Telekom’s own solution, ‘Semasuite®’, can support this.

Using Business GPT safely and efficiently – with government funding. Find out more here.

The UKA Group, a leading renewable energy project developer, is already using Business GPT to optimise internal processes and speed up information retrieval. By deploying an AI-powered chatbot with cloud-based identity and access management, UKA has been able to significantly boost the efficiency of its business processes. The chatbot assists staff in searching for specific information within large volumes of data, drafting texts and consolidating unstructured text. This demonstrates how AI can help improve administrative efficiency whilst meeting high data protection standards. UKA uses the GPT 3.5 and GPT 4 AI models in a secure, PSA-certified cloud environment, which ensures data security. 

Business GPT offers public authorities a secure and sustainable solution to fully harness the benefits of generative artificial intelligence. With high data protection standards, comprehensive security and flexible integration into existing systems, Business GPT enables the automation and optimisation of numerous processes. This leads to significant efficiency gains and cost savings, whilst at the same time improving the quality of service for citizens.

By carefully selecting the right AI solution, public authorities can modernise the way they work and prepare for the future. Business GPT is an excellent choice for safely and successfully navigating the challenges and opportunities of digitalisation.