Wie nutzen Sie künstliche Intelligenz und Analysen, um Ihren Wissensmanagementprozess zu verbessern?

Most platforms (e.g. Helpjuice.com) do this for you, without you having to think, but here's what I'd look out for: AI in KM isn't about replacing humans, it's about amplifying them. We use it for: 1. Auto-tagging and smart categorization. Keeps things consistent. 2. Have AI automatically standardize your articles. You can automate via Zapier or Helpjuice. 3. Draft content generation from conversations and logs. AI's Great at this - we're even working on a tool that does this. 4. Run Analytics reports on AI tools such as ClaudeAI. I'm a bit limited by characters here, but I think you get the drift :-)

I foresee that AI greatly impacts how new information is created, especially for citizen-center government agency digital transformation initiatives. It can automate and improve the process of getting insights from agency operations documents, databases, web pages, social media, and audiovisual information, among other things. Using AI technologies like natural language processing (NLP) and computer vision, writing and speech can be analyzed, and images and videos can be recognized. Machine learning is also very helpful for finding patterns and trends based on citizen interactions with self-service portals.

Utilising AI in the knowledge creation process drastically improves the time it takes to deliver quality knowledge articles. However, that does not mean that you should have a 100% automated process that bypasses human approval. Consider the impact it will have on brand reputation if AI creates something that doesn't understand the human nuance of a particular subject. Aim to use AI to do 80% of the heavy lifting, allowing authors to use 100% of their focus on the remaining 20% of the work to be done (reviewing, ensuring accuracy, approving). The work we have been doing centers around this approach with promising results.

AI can mine data sets, generate summaries and reports, and ensure that knowledge bases are constantly enriched and updated. It can also use NLP to ensure content language is optimized for its intended audience. Using this capability, strategy teams can automatically create summaries of 50+ page documents or groups of documents for presentation. By using productive artificial intelligence, new knowledge can be obtained in various subjects.

Keep in mind that that Chat GPT includes the acronym for Generative Pre-Trained Transformer. AI extracts information on the basis of how it is trained to understand the words and language it reads. For any discipline or field, this means choosing a Gen AI solution that has been properly trained on the language of that discipline itself. For a lawyer, that means that a phrase like "force majeure" is understood for all its complexity by the technology. For a doctor, it may mean that the AI has reviewed thousands upon thousands of images of cancer tumours. The creation of knowledge is predicated on an environment of learning and training- as true a connection for machines as it is for the human species.

You can determine which artifacts are used (and therefore useful) vs. those that sit on the shelf. Then you can make educated evaluations: 1) Are we using the right artifacts? 2) Are the artifacts that are ignored still useful? 3) Is the structure of our repository creating bias in the way people utilize artifacts? Answers to these questions can inform your move-forward strategy.

Let's say you have a large database of customer interactions. By applying analytics, you can identify trends in customer preferences, common issues, or emerging needs. For example, analytics might reveal that customers are increasingly interested in a specific product feature, prompting your team to focus on its development. By leveraging analytics tools such as data mining or natural language processing, you can extract meaningful information from vast datasets, helping your organization make informed decisions and stay ahead in the market.

Indeed, there may be a slight variation to the approach. Envision having access to a repository of knowledge from your colleagues and a vast resource on the internet. By identifying the gap between your lessons learned and the information available online, you can potentially generate novel topics for exploration and discovery. For instance, if you provide information on how to repair a specific tool, the AI can comprehend this knowledge. Simultaneously, the AI can locate a lesson learned about a similar tool and establish comparisons between the two. Leveraging this capability, the AI can propose innovative solutions to enhance your experience and contribute to the creation of new lessons learned.

It's important to note that IT, the Internet, and even "knowledge management" the way it is currently understood in computing is in a very nascent stage. Humans arrange knowledge not just by words whether placed in physical form (books) or wikis, etc. Knowledge management and organization first started with arranging symbols -both graphical and in physical 3D -- in spatial contexts. I suspect even before language of the written form (which is around 5000-6000 years back) humans used spatial organization to handle knowledge. And still do. One need to research deeply into how architecture has played a crucial part in knowledge management before making a statement that it would be databases, wikis, etc. that would yield knowledge!

Incorporating AI and data analytics into your KM ecosystem can be a game-changer. For example, AI can fine-tune knowledge delivery, like suggesting personalized articles for employees. Predictive analytics might forecast industry shifts, guiding strategic decisions. This integration streamlines operations, cuts costs, and sparks innovation, keeping your business at the forefront. Embrace these tools to transform knowledge into a catalyst for growth and a competitive advantage.

Let’s explore how AI and analytics can boost knowledge management: - Smart Tagging: AI automates content organization, simplifying search. - Personalized Suggestions: AI recommends relevant content based on user behavior, like Netflix suggests shows. - Language Processing: AI powers chatbots for interactive knowledge access. - Predictive Insights: Analytics anticipate trends, aiding proactive content creation. - Sentiment Analysis: AI gauges user feedback, ensuring a positive experience. - Feedback Loop: Analytics refine systems based on user interactions, improving over time.

I have recently seen a few awesome examples of GenAI assisting (medical) experts in delivering knowledge. The expert makes a very short analysis and conclusion including next steps to take. The GenAI tool then drafts both the internal requests to other departments (for analysis etc) and a more elaborate and considerate letter to the patient. Combining different writing styles and producing those letters almost instantly, based on the analysis and conclusion by the expert is a huge time-saver!

I hope for AI & analytics help to address KM challenges: leverage large volumes of data, flag misinformation, maintain security... but I don't think we are there yet. 1- Sources, domain and formats (text/image/A/V/structured /unstructured/combined languages) need multiple models. 2-We expect a magic wand to do what we need. But users have to know and express their requirements (questions, security constraints...) 3- The belief in AI can be so strong that the human effort/input is forgotten in both (a) the design and contextualization of the tool and (b) the validation of its contribution. The cost of training & adjusting the tools is not budgeted and poor results are either blamed to humans or the tool is just abandoned as a bad product.

Much of AI sneakily use reductionist principles -- some methodologies are actually so fraught with logical errors that detected patterns can be easily fooled. The question; "what defines the boundaries of a knowledge silo" is still not fully answerable. AI, as it is currently seen in machine learning goes about "dividing and conquering" complex inter-connected holistic in biased chunks. That is one of the reason why efforts such as ChatGPT etc can blithely lie. Good understanding cannot come from reductionism. It requires both reductionist and holistic approaches. Unless AI sees a proper dance that examines both the analytical and the synthetical parts of the subject being studied, it is actually fraught with a lot of "silo-ing" errors

Ensuring accurate and relevant AI-generated insights requires combining AI with human expertise. Address data privacy concerns by adopting robust security measures and compliance protocols. Streamline AI integration by training staff and leveraging user-friendly AI tools. Regularly evaluate AI performance to fine-tune algorithms and optimize knowledge management processes.

AI can help to efficiently reduce the quantity and diversity of documented knowledge. However, the successful reduction of knowledge silos requires the commitment and willingness of everyone involved. AI systems are based on the data with which they are trained. Without careful curation, they run the risk of using inaccurate or irrelevant information, which can lead to inaccurate results. Curation ensures that AI works with high-quality, consistent and relevant data, which improves its efficiency and effectiveness. Employee expertise and tacit knowledge as well as a culture of open knowledge sharing are crucial to ensure the quality and benefits of knowledge management and to guarantee the successful use of AI in organisations.

Think about how Netflix suggests shows based on what you've watched - that's AI at play, learning your preferences. Similarly, in business, AI-powered tools can analyze user actions to predict what information employees need, saving time and effort. Analytics add another layer by crunching data to uncover hidden patterns, guiding smarter decisions. Just like Spotify adjusts playlists to suit your mood, AI tailors knowledge delivery to fit each user's context. Over time, it gets even better, continuously learning and refining its suggestions. This dynamic duo doesn't just organize knowledge; they unlock insights and boost productivity, giving businesses a competitive edge.

Developing a collection of useful "AI prompts" is low-hanging but important best practice initiative for any organization adopting AI solutions. One aspect of KM will never change - the need for human oversight to avoid "garbage in, garbage out". This was fundamental to search engine developments and will now be equally important to AI programs. Don't make the assumption that any new technology is a complete self-starter. This technology needs cultivation and nurturing and there is no better place to start than "prompting" libraries. Peter Druker, perhaps the greatest thinker on management consultant, said "the most serious mistakes are not being made as a result of wrong answers. The true dangerous thing is asking the wrong question".

AI chatbots can handle routine queries, freeing up human resources for more complex tasks. Analytics tools can track user interactions with knowledge platforms, highlighting popular topics or areas needing improvement. Utilize AI-driven recommendation engines to suggest relevant content based on user preferences, enhancing user experience. Regularly assess AI performance and refine algorithms to ensure accuracy and relevance. By integrating AI and analytics seamlessly into KM processes, you can boost efficiency, enhance user satisfaction, and derive valuable insights for continuous improvement.

Nowadays, many organizations are at an early stage of KM maturity. In these cases, it’s important to carefully analyze the real benefits of applying AI. Sometimes, organizations 'fall in love' with new technologies or trends and rush into implementation. When this happens, results may fall short of expectations, or budgets may not be sufficient, leading to KM activities being deprioritized because they are seen as 'unnecessary.' This way, organizations miss the opportunity to strengthen processes and fully realize the benefits of KM

Train the trainer! It is essential to train users, stakeholders and ourselves in the selection, use, and integration of AI tools by using AI tools. Our experience is crucial for developing purpose-oriented ideas for the effective use of AI. Take it as a continuous feedback loop. We need our expertise as a reference to be able to make a well-founded assessment of the results of AI use based on our own use cases.

GenAI can be a quick and powerful tool. But a lot of the applications of GenAI can much better be done by good-old statistical (regression / correlation / etc) analysis. GenAI will pull you to the middle; the most common phrases and conclusions will come first and you risk following the majority. When succes in the future is different than success in the past, don't use models trained on the past (which are all GPT models, by definition). Amazon learned this the hard way when classifying incoming CVs based on previously successful hires; to their surprise only white males were selected.

I have worked a lot in understanding information and knowledge organization in architecture. In my field, BIM (Building Information Modelling) already shows a lot of bias towards preserving and uplifting commercial agendas instead of supporting the wider ramifications of architecture. I suspect even other specializations such as medicine, etc. would also see serious biases when knowledge is based on AI or predominantly on it. There is a huge lot that still needs to be done in AI. We should not be carried away by the hype "valley" we are all inside.