Suggested prompts help users avoid Blank Canvas Syndrom.
Sample suggestions help a user learn what they could ask the system to do, and keep the generative conversation moving forward as it progresses. Generally these appear in a list of 3-5 suggestions that pre-fill the chat input when selected.
This pattern is very similar to templates and nudges, other wayfinding devices. The difference here is that these are generally used in the open chat request type to kickstart or a continue a conversation, whereas these other patterns are actions or automations can make it easier to interact with the AI, particularly at first.
Getting started
Suggestions can show you how to get started, but without context for the user's needs or intent, they are more likely to be irrelevant than not. In fact, an internet joke is already forming around the insistence of AI companies to let their chatbot plan your vacation.
For example, I just started a new ChatGPT conversation. Despite months of use at the premium tier, the four options it provides to me are completely irrelevant to my interests and my work:
Without personalization and context, suggestions have negligible use for returning users. As the systems learn the users through personalization and memory, they are more likely to become useful on subsequent visits as a quick way of getting started.
For example, imagine you write the same 4 commands into the terminal every time to start a coding project. If Github's copilot knows this, it can combine a prompt to kickstart that workflow into a single suggestion that stakes just a click to fire off.
As follow ups
Suggestions are most helpful when used mid-conversation. In this case, they can include contextual information and make the act of interacting with an AI chatbot or copilot far less laborious, while still giving the user control to command the conversation.
For example, Github Copilot automatically updates its suggestions as the conversation proceeds. It's suggestions can be quite long, reducing the time it would take a user to ask a common question from several seconds to a single click.
Meta, LinkedIn, and other feed-based products have increasingly added suggestions below individual posts as well, though the quality and relevance is hit or miss. This may lead you to consider whether the incentive here is really to increase engagement with (and the flow of training data into) their proprietary AI.
