Right now, it’s hard not to notice the hype around corporate artificial intelligence (AI) use caused by ChatGPT, a generative AI tool. At one end of the spectrum, it can provide better answers than Google (and by “better,” I mean more focused and more quickly than taking the Google route) and, at the other, there are opportunities to embed ChatGPT-based capabilities into IT service management (ITSM) and enterprise service management (ESM) operations.

This blog shares what ChatGPT is and how it works, the pros and cons of using ChatGPT, and examples of how it can be employed in ITSM and ESM operations to improve operations and outcomes.

This @SysAid blog shares what #ChatGPT is and how it works, the pros and cons of using ChatGPT, and examples of how it can be employed in #ITSM and #ESM operations to improve operations and outcomes. #AI #ServiceDesk Click To Tweet

How ChatGPT works

ChatGPT is an advanced natural language processing (NLP) model based on the OpenAI Generative Pre-trained Transformer (GPT) architecture. It’s a generative model, which means it generates text by responding to a prompt or answering questions in a conversational manner. Importantly, it doesn’t have a fixed set of “canned” responses and creates its answers dynamically based on the input it receives.

The underlying ChatGPT architecture is based on transformers – neural network architectures that are good at handling sequential data and capturing dependencies and relationships in the text. At its core, ChatGPT is a language model, which means it is trained to predict the likelihood of a sequence of words appearing in natural language. It uses this ability to generate grammatically correct and contextually appropriate text.

The model is pre-trained on a massive body of text data from the internet, allowing it to learn language structure, including grammar, facts, and reasoning abilities. ChatGPT discovers the statistical properties of the language and generates a representation of words, phrases, and language constructs. This learning is unsupervised.

After the pre-training, ChatGPT is fine-tuned on a more specific dataset to make it more adept at engaging in interactive dialogues. This phase involves training the model on a narrower dataset with human feedback.

ChatGPT can be used in various applications, including chatbots, content creation, language translation, and information retrieval. ChatGPT uses the learned statistical properties and relationships to generate a response when a text prompt is provided. The model considers the context provided by the input and uses its knowledge and training to compose an output that is likely to follow the input text based on the patterns and information it has seen during pre-training.

Finally, it’s important to appreciate that ChatGPT doesn’t understand the text. Instead, it generates responses based on the patterns and associations it learned during training. Top of Form

Recognize the pros and cons of ChatGPT

Before using ChatGPT to enhance your ITSM and ESM capabilities, knowing its pros and cons is essential. Two limitations are called out in the tool’s web interface:

1. ChatGPT’s last knowledge update was in September 2021, i.e. its answers don’t include anything newer than this. For some ITSM use cases in particular, this knowledge gap will cause issues.
2. The statement “ChatGPT may produce inaccurate information about people, places, or facts” is visible below the prompt box. This second limitation is different from the first, and the relative importance of the ITSM or ESM ChatGPT use case might require human checking of ChatGPT’s suitability.

There are many positives, though. These pros include that ChatGPT can:

• Handle a large volume of queries and tasks and provide 24/7 service. Such scalability can be problematic without intelligent automation in terms of human availability and costs.
• Be more cost-effective than manual operations, especially for repetitive and high-volume tasks.
• Be trained to understand and respond in multiple languages, making it useful for global communication and services.
• Process and provide information quickly. It can also help in summarizing large datasets or documents efficiently.

In addition to the two acknowledged limitations, other cons must be considered when assessing the use of ChatGPT in ITSM and ESM capabilities. These potential negatives include that ChatGPT:

• Doesn’t understand language or context in the same way humans do. Instead, it generates responses based on the patterns in the data it was trained on. This approach can lead to incorrect or nonsensical responses.
• Can inadvertently learn any biases present in the training data.
• Responses are dependent on input phrasing. Such that responses can be inconsistent based on slight changes to the input phrasing.
• Can be overly verbose or reuse certain phrases excessively, which might prevent concise and effective communication. For example, unnecessarily long or complicated knowledge articles are unhelpful to end-users.

Before using ChatGPT to enhance your ITSM and ESM capabilities, knowing its pros and cons is essential. This blog by @SysAid explores. #AI #ArtificialIntelligence #ChatGPT #ITSM #ServiceDesk Click To Tweet

Example ChatGPT ITSM and ESM use cases

ChatGPT can be employed in various aspects of ITSM, and ESM as an extension of this, to improve service and support operations and outcomes. Examples include:

• Incident management – automating ticket triage and providing initial diagnostics, suggesting potential solutions or workarounds.
• Self-service – ChatGPT can be used as a chatbot and to retrieve relevant articles from the knowledge base.
• Problem management – analyzing incident data to identify recurring issues or patterns, analyzing incident and log data to identify potential root causes, and predicting incidents before they occur.
• Change enablement – assessing the impact and risk of proposed changes by analyzing historical data.
• Service configuration management – automating configuration item (CI) updates in the configuration management database (CMDB) and identifying and mapping dependencies between different CIs.
• Knowledge management – curating and maintaining the knowledge base and improving information retrieval.
• Service level management and continual improvement – monitoring service levels and generating reports to help ensure service level agreement (SLA) targets are met, sending automated alerts and notifications regarding potential SLA breaches, analyzing customer feedback for service quality and end-user satisfaction insights, and suggesting areas for service improvement.

ChatGPT can be employed in various aspects of ITSM, and ESM as an extension of this, to improve service and support operations and outcomes. Take a look at these 7 examples via @SysAid. #ChatGPT #ITSM #ESM #AI #ServiceDesk Click To Tweet

If you would like to learn more about the many ChatGPT use cases in ITSM and ESM, please get in touch.