Zum Inhalt springen
YouTube page opens in new windowFacebook page opens in new windowLinkedin page opens in new windowInstagram page opens in new windowRSS page opens in new window
European Digital Innovation Hub Saarland
European Digital Innovation Hub SaarlandEuropean Digital Innovation Hub Saarland
  • Termine & Veranstaltungen
  • Angebote
    • Schulungen
    • Förderprogramme
    • Best Practice
    • Webinar
    • Technologieradar
  • News
  • Über uns
  • de_DEDeutsch
    • en_GBEnglish (UK)
    • fr_FRFrançais
  • Termine & Veranstaltungen
  • Angebote
    • Schulungen
    • Förderprogramme
    • Best Practice
    • Webinar
    • Technologieradar
  • News
  • Über uns
  • de_DEDeutsch
    • en_GBEnglish (UK)
    • fr_FRFrançais
Juli302024

Chat GPTs as knowledge management?

In today’s knowledge economy, companies rely on efficient knowledge management systems.
Large language models (LLMs) offer an innovative solution for managing and utilizing internal knowledge.
This article examines various approaches to implementing LLMs in the corporate context as well as the opportunities and challenges they present.   A key problem for many companies is the shortage of skilled workers.
When experienced employees retire or leave the company, valuable knowledge is lost.
LLMs can help to document this knowledge and make it accessible to other employees.
A well thought-out knowledge management system can thus help to secure and promote the transfer of knowledge within the company.
Another important aspect when implementing LLMs is cyber security.
Comprehensive security measures must be taken to protect sensitive information.
These include the encryption of data during transmission and storage, strict access controls and regular security audits to identify and rectify vulnerabilities.
In addition, it must be clearly defined which data may be used.
Sensitive and personal data should only be used in accordance with the applicable data protection regulations.   The use of LLMs in knowledge management is still in its infancy, but the potential is enormous.
Future developments could include more powerful and efficient models as well as hybrid systems that combine the strengths of humans and AI.
In addition, LLMs could be used in more and more areas, from customer service to product development.

The usual suspects are there too, of course… Providers such as OpenAI, Microsoft Azure or Google Cloud offer powerful LLM-based solutions that can be implemented quickly and easily.
These services are highly scalable and provide access to the latest models and technologies.
However, the use of external service providers also involves risks, particularly in the area of data security and compliance.
The transfer and storage of data outside your own infrastructure can present potential vulnerabilities.
In addition, the running costs for cloud services can be higher in the long term.  

Local hosting prevents the data from leaving the company’s infrastructure. One option is to operate LLMs on the company’s own infrastructure.
This method has the advantage that sensitive data does not leave the company network, which is particularly important in data-sensitive industries.
The data remains under the full control of the company, which makes it easier to comply with data protection guidelines.
Security can also be ensured through tailored measures and stricter access controls.
However, this approach also comes with significant challenges.
The cost of hardware and its maintenance is usually high.
In addition, the implementation and operation of such systems require special knowledge and resources that smaller companies in particular are often unable to provide.
Scalability can also be problematic, especially when the volume of data and the number of users increase.
One example of a local implementation is the use of Retrieval Augmented Generation (RAG).
This technique combines the strengths of retrieval systems and generative models by extracting relevant information from large data sets and integrating it into the generation of answers.  

If the data is only to remain in the country… Hosting the data on a cloud service in your own country can be a middle way.
This solution is particularly useful if the internal cyber security standards are not sufficient or the resources are lacking.
Hosting with a trustworthy provider in your own country offers a balance between security and practicability.
Professional providers often have better security measures in place and compliance with national data protection laws is easier to ensure.
However, there are also more and more innovative companies that offer all-in-one systems with interfaces to most common data sources and hosting in Germany.

What to do if employees don’t want to use the system? Another key to the success of a knowledge management system is employee commitment.
The best system is useless if employees do not use it.
It is therefore important to motivate and involve them.
Approaches such as gamification, where rewards and challenges encourage use, can be helpful here.
Regular training and workshops as well as the inclusion of employee feedback in the further development of the system also contribute to acceptance.  The right system and acceptance within the team determine success In summary, it can be said that LLMs offer a promising opportunity to revolutionize knowledge management in companies.
The choice between local and hosted solutions depends on individual requirements and circumstances.
Data security and data protection remain key challenges that need to be addressed carefully.
Against the backdrop of skills shortages and knowledge loss, LLMs offer a valuable solution for securing and sharing corporate knowledge.
Strategies such as gamification and user-friendly design promote long-term use by employees.
When it comes to implementation, professional advice is crucial to exploit the full potential of the technology while minimizing the risks.  Contact EDIH Saarland, we will support you Arrange an individual consultation appointment with our experts.
We will help you find your way through the jungle of solutions and define the right questions for your problem.

Make an appointment for a consultation today

Author

M. Sc. Marco Giangreco
Digitization expert with a focus on AI
Mai302024

AI in everyday life: advisor, personal assistant and bodyguard

Artificial intelligence (AI) is currently on everyone’s lips. If you had talked about this topic ten years ago and listed examples, most people would have thought of self-driving cars and robots. But the possible applications in everyday life are also almost unlimited. In fact, we claim that you’ve already used 5 AI technologies this week – perhaps without even realizing it. How many points on this checklist do you recognize from your everyday life?

 

AI as a cell phone guard: the face as a password

A quick glance at the camera and the cell phone is unlocked. Thanks to intelligent facial recognition, the latest smartphones can be unlocked without a PIN or fingerprint. Even in poor lighting conditions, with a different hairstyle or new glasses. This function is based on AI, because when facial recognition is set up, a detailed image of the face is captured via the front camera and analyzed and stored using complex algorithms. With each unlocking process, the stored analysis data is compared and further training data is collected. Some cell phone models also assign the pictures in your gallery to specific people or animals, for example, so that you have all the photos of your favorite Labrador mix Polly at your fingertips with just one click. This feature is also based on AI.

 

AI as a personal assistant: devices that obey your every word

A short command and the light is switched off, a call is made or the next route is selected in the navigation system. AI voice recognition makes it possible to control smart speakers, cell phones or cars by voice command. Natural language processing (NLP) plays a key role here, as it is necessary to recognize what the person has said (Automatic Speech Recognition) and what they want to achieve with their statement (Natural Language Understanding). If these blocks are successfully recognized and merged, the request is processed and the command is executed.

 

AI as the best advisor: recommendations just for you

Just one click and you’ll receive tailored recommendations when shopping online, in your podcast app or from your favorite streaming service. AI is increasingly being used for personalized recommendations and advertising. Based on data collections and analyses, so-called user profiles are created using machine learning, on the basis of which recommendations are generated that best match the user’s individual preferences and interests. The advantages for providers and consumers are manifold: companies can offer a customized consumer experience that sets them apart from the crowd and can lead to an increase in sales. At the same time, potential customers are more satisfied because they find the right product more quickly – and sometimes they even find things they didn’t even know they needed 😉

 

AI as a mediator: language talent, thanks to AI

One click and you are speaking Chinese. There are around 7000 languages in the world – and just 2% of Germans have knowledge of 5 languages or more.[1 ] We often find ourselves in situations where language barriers can lead to limited access to information or misunderstandings – and sometimes we are just proud to be able to order our food on vacation in the language of the locals. Machine translation (MT) has become an integral part of many people’s everyday lives, as it not only enables efficient communication across borders, but also multiplies the availability of information and thus access to knowledge and resources. With MT, text is automatically transferred from one language to another. Neural machine translation marked a real breakthrough, as the systems are trained with the help of self-learning algorithms, large amounts of data and immense computer capacities. Instead, all text elements are analyzed and it is recognized how the words influence each other and how they relate to each other. Nevertheless, it must be said: Even if entire manuals or product reviews are now completely machine-translated and you don’t even notice it at first glance with some texts, not every text is equally suitable for MT.

 

AI as a bodyguard: no chance for spam mails in your inbox

Is your spam folder full even though you haven’t filled it? AI has probably done that for you. By using AI-based methods, email providers can filter spam more effectively and reduce the amount of unwanted emails. Incoming emails are scanned using machine learning and checked for warning signals such as malicious IP addresses and links, suspicious attachments or suspicious keywords. If enough criteria are met, the message is marked as spam and moved accordingly. If the marking is incorrect or a spam mail sneaks past the AI bodyguard, this error can be corrected manually by marking the message accordingly. This will help the system to learn and adjust its parameters, which will further train and improve it.

 

These are just a few of many examples that show that AI is for everyone! If you would like to gain further exciting insights into innovations, AI and virtual reality and experience for yourself how they can strengthen competitiveness, visit us at the 15. and June 16 at our EDIH Saarland stand at the
make-it.saarland 2024
at the E-Werk Saarbrücken.

[1] https://de.statista.com/statistik/daten/studie/1267264/umfrage/umfrage-zur-anzahl-gesprochener-sprachen-in-deutschland/#:~:%20mehr%.

Make an appointment for a consultation today

Author

M.A. Eileen Marra

Researcher & Manager of EDIH Networking Office

European Digital Innovation Hub Saarland
  • Adresse
    c/o ZeMA, Eschberger Weg 46, D-66121 Saarbrücken
  • Telefon
    +49 (0) 681 85787 – 300
  • E-Mail
    info@edih-saarland.de

Der European Digital Innovation Hub Saarland (EDIH Saarland) wird mit bis zu 50% aus EU-Mitteln gefördert (GA 101083337) sowie vom saarländischen Ministerium für Wirtschaft, Innovation, Digitales und Energie. Der EDIH Saarland bietet den KMUs in der Region einen kostenlosen One-Stop-Shop für Digitalisierung und Anwendung von Künstlicher Intelligenz (KI). In den nächsten drei Jahren (2023-2025) wird maßgebliche Expertise für den Praxistransfer von industrieller KI im Saarland, der Großregion (Saar-Lor-Lux) und in Europa bereitgestellt.

Das ZeMA ist hier federführend, neben den beteiligten Projektpartnern AWSi, DFKI, saaris und East Side Fab.

European Digital Innovation Hub Saarland
  • Impressum
  • Datenschutz
  • Privatsphäre-Einstellungen ändern
  • Historie der Privatsphäre-Einstellungen
  • Einwilligungen widerrufen
Rechtliche Hinweise & Cookie-Einstellungen

© European Digital Innovation Hub Saarland

Go to Top
Cookie Consent mit Real Cookie Banner