The future of Knowledge Management in the time of Generative AI
The rise of generative AI like ChatGPT, has the potential to revolutionize the way we manage and share knowledge. It can transform the way organizations handle their knowledge management processes, leading to increased efficiency, accuracy, and personalization. However, it also presents new challenges and risks that must be taken into consideration before implementation. At the end of the day ChatGPT is Our Latest Collaborator, Not a Job-Stealer.
Benefits of using generative AI in knowledge management
Generative AI can bring huge benefits to the world of knowledge management, including but not all-inclusive:
- Quick and efficient generation of new information: Generative AI can quickly generate new information and insights based on existing data. In a KCS world with structured case notes and article templates, this becomes even easier to kickstart article creation for editing by SMEs before publishing.
- Earlier resurfacing of known information: Imagine a support world where a customer is provided the right content within the first two steps of their interactions when seeking a solution to an issue. We know that 70%-80% of support interactions are based around issues that have been previously solved. When Generative AI is used to assist, we could see an enormous reduction of support cases generated, leaving our teams to focus on the more complex and interesting issues.
- Increased accuracy and reliability of information: With local/private generative AI instances, organizations can ensure that the information they provide to customers and employees is consistent, accurate, reliable, and up-to-date. This can help build trust and improve customer satisfaction.
- Improved ability to analyze and make sense of large amounts of data: With generative AI, organizations can more quickly analyze vast amounts of data, identify patterns, and make sense of complex information. This brings deep efficiencies to the analytics realm, where finding patterns faster means more time can be spent analyzing the stories behind them. This can help them make better-informed decisions and of course gain a competitive advantage.
- Increased automation and efficiency in knowledge management processes: Generative AI can automate repetitive tasks, such as data entry and analysis, freeing up employees to focus on more important tasks. This directly leads to increased efficiency and productivity in task-based roles.
- Improved ability to personalize knowledge management based on individual user needs: Generative AI can provide tailored information and insights that meet specific user requirements. By leveraging our “know your customer” type initiatives around personalization, we can deliver the right information without guessing, which can help improve customer satisfaction and engagement when our systems remember what they’ve told us previously. This leans heavily into the ongoing work around the Predictive Customer Engagement model.
Risks associated with using generative AI in knowledge management
While generative AI can bring many benefits to the knowledge management world, it also presents new risks and challenges that organizations need to consider. These risks include:
- The potential for inaccurate or misleading information to be generated: Generative AI can generate new information that has not been previously reviewed, which can lead to inaccurate or misleading information being shared with customers and employees. The easy mitigation here is to use AI to draft content for your SMEs to review before publishing.
- The risk of overwhelming users with too much information: Generative AI can generate vast amounts of data, which can be overwhelming for users and make it difficult for them to find the information they need. Healthy content lifecycle management strategies are critical here to ensure relevant knowledge is more readily discovered. (Book a call to ask me how I can help you with that challenge!)
- The potential for generative AI to be used unethically or inappropriately: Generative AI can be used to create content that is unethical or inappropriate, which can damage an organization’s reputation and lead to legal issues. Again, no direct publishing is a critical piece here as is a solid content governance process and clear style guide definition. But, while this is a risk, it is also an area where AI can help solve the issues as well, like with Acrolinx’ AI Enrich just recently released.
- The risk of privacy violations or other security breaches: When not trained correctly, generative AI can expose sensitive information. This is a gold-mine target for hackers and other malicious actors. Organizations must take measures to protect this information from unauthorized access. This includes understanding the necessary training needs for your LLM to restrict PII from being delivered, but also means ensuring company IP is handled appropriately as well. Training solely based on publicly available information in your knowledge base is the safest method and why a *KCS 90/0 rule can be so impactful to your success here. (*90% of what we know is published externally at or before case closure.)
Use cases* for generative AI in knowledge management
There are soooo many potential use cases for generative AI in the knowledge management world, not the least of which are:
- Customer engagement and satisfaction: Generative AI can be used to provide personalized, relevant, and fast information to customers, shortening their time to resolution thus improving their engagement and satisfaction.
- Product improvements: AI can be used to identify areas for product improvements based on content trends and patterns, as well as customer feedback and usage data.
- Automating audits, classifications, and summaries: AI can automate the process of auditing and classifying content, as well as summarizing information for easier consumption and analysis.
- Finding and repairing knowledge gaps and duplicates: Generative AI can identify knowledge gaps and, even more importantly, duplicates in a knowledge management system, improving its accuracy, discoverability, and reliability.
- Responding to support requests and triaging issues: AI can easily assist in responding to support requests of known issues and and then triaging new issues, freeing up employees to focus on those more complex and interesting cases.
- Personalizing search results and assistance: We touched on this one above already, but generative AI can personalize search results and assistance based on individual user needs (think web account profiles, CRM data, and past case histories), improving their experience.
- Delivering coaching and training materials: This is a big one that alone has its own pros and cons attached to it… AI can be used to deliver coaching and training materials to employees and/or kickstarting coaching conversations by recommending topics for discussion, improving their skills and knowledge. More on this can be found in this Forbes article recently published by Anna Gallotti: Pros and Cons of Coaching with AI.
- Predictive analytics for customer temperature and escalations: Generative AI can be used to predict customer temperature and escalations, allowing organizations to proactively address issues before they become problems. With our heavy shift to SaaS implementations, this is becoming even more necessary and simultaneously easier to accomplish at scale.
- Providing leadership-level summaries of customer cases: Has your manager, director, VP or C-suite exec ever asked to be read-in on a customer issue? Generative AI can provide leadership with summaries of customer cases, allowing them to see the bigger picture and make better-informed decisions.
- Generating summaries and translations for knowledge articles and so much more: This is of course one we all know by now; AI can generate summaries of knowledge articles, lengthy blog posts (like this one), even Youtube videos and podcasts, and then translate them to various languages thereby improving their accessibility and readability to broader audiences.
- Suggesting responses to community questions: Community sourced solutions are a goldmine of knowledge and generative AI can both learn from as well as suggest responses to community questions. Meeting your customers where they are just got a bit easier thanks to AI, and you reap the benefits from the ability to collate all that knowledge into your codified systems of record as well!
- Creating sample code and troubleshooting scripts: We’ve seen this a lot recently as well, generative AI can create sample code and troubleshooting scripts, improving the time it may take for an engineer to accomplish the same task starting from zero. Using this capability can add deep productivity and efficiency gains to your support staff nearly immediately.
- Providing support site assistance and assistant tools: This one has already been in use for so long we likely have forgotten about how Generative AI can provide support site assistance and assistant tools rather than just being a ‘dumb’ chatbot that only knows how to respond in predefined Q/A pairs. By using the generative side to AI, you can improve the overall customer experience by giving them a much more robust conversational bot that provides answers not just a small subset of very specific task-based capabilities.
*You can even get your very own “What’s Your Use Case?” t-shirt from the Consortium for Service Innovation’s Zazzle store!
Considerations when implementing generative AI in knowledge management
All of these really lead us now to the things we need to consider deeply when beginning to work with generative AI in the knowledge management world:
- The need for rigorous oversight and management of generative AI systems: Organizations must carefully manage and monitor generative AI systems to ensure that they are generating accurate and reliable information. Starting with a clear sense of governance needs will go far in your success here.
- The importance of having effective knowledge management systems in place to ensure accuracy and reliability of information: We MUST have effective knowledge management systems in place to ensure that the information generated by generative AI is accurate and reliable. Clean up your own house before training an LLM on your knowledge base. Junk in, junk out as they say.
- The need for clear guidelines and training on the ethical use of generative AI: Organizations must establish clear guidelines and provide training on the ethical use of generative AI to prevent unethical or inappropriate use. Many companies and countries are wrestling with building these guidelines today. Build your own or repurpose another, but be certain those guidelines are in place before rolling out your AI solution.
- The importance of privacy and security in managing sensitive information: Organizations must take measures to protect sensitive information generated by generative AI from unauthorized access. This is likely the biggest global conversation around AI today. Jumping into public LLMs like ChatGPT can quickly find you out of privacy and security compliance and into scary litigious territories. Not to mention accidental exposure of company confidential intellectual property. Stay safe my friends and make sure your LLM is trained on the right data, not all of your data. Looking to hosting solutions inside your firewall with models you control is a great start to this. Checkout Huggingface.co to see how may models are available to being playing with and even replicate github repos into your own ecosystem to begin some investigative work.
- The potential for generative AI to complement, rather than replace, human knowledge management expertise: Organizations must recognize that generative AI should be used to complement human expertise, rather than replace it entirely. The biggest fear in the workforce today is that AI is going to take their jobs. Without appropriate recognition and consideration by the executive suite this could be one ramification. But fret not, if we can adapt (like we always have throughout the history of humanity, and especially through the information age), we can use the tools at hand to improve our work, not take it away…. which also leads me to…
What Knowledge Management does that AI cannot yet accomplish
We take tacit knowledge and make it explicit knowledge:
We, as learned knowledge managers, know things (and drink, sometimes), but often we miss codifying that knowledge into systems of record. There’s no AI in the world (yet) that can take what we’ve learned through our years and put it into any structured source. We are the ones explicitly tasked with sharing our knowledge in discoverable and consumable ways. Otherwise, this institutional knowledge is lost to the winds of time and must be discovered anew each time it is warranted leading to slow progress in any discipline.
We capture what we think we know:
Learning stems from sharing what we know, even if we aren’t certain, by processing feedback that challenges our ideas with facts, and then reforming our ideas once again. Nothing can be challenged nor validated if it isn’t shared. In the corporate world, and especially in the KCS spaces, this means documenting what we think we know, and allowing others to use that information. Which leads directly into….
We reuse to review:
Only when we reuse that shared knowledge can we validate or improve upon it. Admittedly, this is one area where AI will surely shine as the LLMs grow ever smarter through training. But it is only through the human review that we are able to train these models, thus, for now proving that KM wins out in this arena as well.
We ensure the tool serves our processes; our processes do not serve the tool:
In a recent Consortium for Service Innovation meeting, Christina Roosen of Akamai says: “Should is not a use case”, and in that pithy sentence we understand exactly that our tools need to fit into our larger goals and purpose; to have a specific use designed to achieve those goals. When we shoe-horn tools and adapt processes to accommodate, we begin to lose sight of what we’re really trying to accomplish.
We help Individuals become recognized as leaders:
Our own individual eminence in our fields is only able to be recognized when share our knowledge. A knowledge sharing culture is where we shine as subject matter experts and begin to really grow in our roles, careers, and lives beyond work. No AI is going to raise you to any level of recognition until you begin sharing your knowledge as a way to self-advocate, but even more importantly, to help others.
The rise of generative AI presents both opportunities and challenges for the knowledge management world. By implementing effective and ethical knowledge management systems and considering the potential risks and benefits, organizations can harness the power of generative AI to improve efficiency, accuracy, and personalization in knowledge management processes. It will require us to adapt. Our jobs will not be the same as they are today. But, generative AI is not going to steal our jobs either. It is going to make it easier for us to do the mundane repetitive or time consuming parts, giving us more opportunity to dig into the deeper parts of our jobs we simply didn’t have time or energy for previously. This is where progress is discovered, and where we are able to build on existing knowledge to further our understanding instead of discovering the same bits over and over again…
Knowledge Management in the time of generative AI is an exciting space to work in, and one that is even more important than ever.