Onboarding

There's a variety of interactions that can help users interact with AI in a clearer and more natural manner. Incorporate different patterns to help users understand what type of inputs are best and communicate them in an easy manner.

Smooth transitions

Smooth and continuous animations can help signal the AI is still processing, and helps users wait out loading times.

Chat GPT’s chat feature shows a continuing circle animation to help users know that the answer is not complete.

Mozilla’s chat interface shows answers are complete through an explicit message and its animated, loading response.

Keep users engaged

Depending on the model, AI responses can be slow. Give your users something to read or keep them engaged with status updates while the response is loading. Share status updates and be transparent about what is happening behind the scenes, such as Midjourney’s fuzzy, loading image states. This allows users to better understand what’s going on under the hood, especially if it’s taking a while. Be careful to avoid any jargon or technical terms.

In addition to an animated typing indicator, Ash’s AI assistant shows a “thinking” state while its answer is being generated.

Luma progressively increases the fidelity of its results as it completes its results.

Avoid text prompts

Provide manual controls so users can influence the output or they can take over when the AI fails. Where possible, avoid having the user enter long and complicated textual prompts. Turn frequent prompts into easily accessible shortcuts that show up when relevant.

Perplexity shows some recent, popular prompts to help users decide what to search.

Microsoft Copilot in Bing helps users learn more by providing prompts relevant to their last question.

Next guideline

Error handling