Tasks of AI Services

AI services can solve diferent tasks. The task is usually part of the serviceId to classify the expected result of the service.

A list of implemented service connectors/features: Services

The following tasks are currently supported ...

chat

A chat task generates conversational text that is relevant, coherent and knowledgable given a prompt. ChatGPT is the probably the most famous chat AI service that provides a chat task.

In the near future, such services can also input and output images, audios and videos beside plain text.

Example use cases:

• chatbot: to have conversations; e.g. used to provide customer service or sales
• assistants: help the user with personal or business tasks
• text generation: create properly worded text snippets for conversions (e.g. emails) or digital content platforms (e.g. social media posts, online shops)

More information and inspiration for chat tasks:

• Example Text summarization, Code completion, and Few-shot learning with chat task of OpenAI
• Example Code Edit with chat task of OpenAI
• The chat equivalent at HuggingFace: conversational
• A chat subset at HuggingFace: question-answering
• Another chat subset at HuggingFace (text-generation)

text-chat

A text-chat task is the same as a chat task, but limited to text inputs and text outputs only.

text-to-image

A text-to-image task generates images from input text. These services can be used to generate and modify images based on text prompts.

Example use cases:

• data visualization: convert data (text) and get image outputs
• concept visualization: conversations can be made more immersive if they provide contextual images based on user text input or based on generated text output
• generation of creative visuals: Different patterns can be generated to obtain unique pieces of visual output, e.g. to make creations easier for designers to conceptualize their design before actually implementing it; useful e.g. in design, fashion, or architecture industry

More information and inspiration for text-to-image tasks:

• text-to-image at HuggingFace

image-to-image

An image-to-image tasks transforms a source image to match the characteristics of a target image, usually controlled by a user text prompt.

Example use cases:

• extend images: fill or replace parts of an image with content in the style of the rest of the image, e.g. to remove unwanted objects or sensor dust
• increase resolution: increase the resolution of an image, allowing for higher quality viewing and printing
• image colorization: black and white images can be brought up to life using an image colorization
• style transfer: Style transfer services can convert a regular photography into a painting in the style of a famous painter.

More information and inspiration for text-to-image tasks:

• image-to-image at HuggingFace

Special Task Identifiers

aisbreaker

An aisbreaker tasks is not a normal AI tasks, instead it's used to address special AIsBreaker services like logging or remote proxy access ...

private Prefix

A task with prefix private (e.g. private, private-chat, private-monster-tool, ...) is application or client specific task with application or client specififc semantics.

Reserved Task Identifiers

All task identifiers not mentioned above are reserved for future use and may not be used at the moment.