Exploring AI Integration and Abstraction

Today we are talking about AI within Drupal, How AI can help, and Modules to use with guest Martin Anderson-Clutz. We’ll also cover Augmentor AI as our module of the week.

For show notes visit: www.talkingDrupal.com/436

• Terminology
• IMF analysis
• Prompt engineering
• AI in Drupal
• Best way to try modules
• Best use of AI
• Other ways of integrating

• Augmentor AI
• Open AI
• Prompt Engineering: Get the Most From Your Drupal Site's AI Integration
• Terminology
  • NLP - work with text provided in a conversational format, understand the intended meaning, and provide a relevant response
  • AI - A subset of CS that aims to develop systems that can mimic human response, or automating sophisticated behavior
  • ML - subset of AI that aims to act without explicit guidance, by extrapolating from known data
  • Deep learning - a subset of ML which uses artificial neural networks with representational learning to develop and leverage their own means of classification and other feature detection
  • LLM - an AI algorithm that uses Deep Learning techniques to accomplish NLP tasks such as responding to unstructured user prompts. LLMs are trained on massive data sets, often gathered from the internet, but sometimes using more specialized data
  • Typically the AI interfaces our listeners are already using are based on an LLM, but the nature and recency of the data they’ve been trained on can vary widely. Recently Mike Miles created Drupal Droid, a GPT model specifically trained for Drupal developers, and you can find a demo of that in our YouTube channel
• Mike Miles Drupal Droid
• AI module list
• OpenAI Image
• Search API Pinecone
• TMGMT Translator OpenAI

Martin Anderson-Clutz - mandclu

Nic Laflin - nLighteneddevelopment.com nicxvan John Picozzi - epam.com johnpicozzi

Martin Anderson-Clutz - mandclu Augmentor AI

• Brief description:
  • Have you ever wanted a highly configurable way to integrate multiple AI services with your Drupal site? There’s a module for that.
• Brief history
  • How old: created in Oct 2022 by murrayw of Morpht, though recent releases are by elonel
  • Versions available: 1.1.2 which works with Drupal 9.5 and 10
• Maintainership
  • Actively maintained, most recent release was earlier this month
  • Documentation available
  • Number of open issues: 11, 3 of which are bugs
• Usage stats:
  • 82 sites
• Module features and usage
  • To use Augmentor AI, you need to define one or more “augmentor” configuration entities. An augmenter entity implements an augmentor type, which determines what AI service it can use, what configuration options it will have available, and so on.
  • The augmentor will define one or more “messages” that provide structure to the prompt that will be passed to the AI model in order to generate a response. It can also be configured in a variety of ways, such as how much randomness to use, a maximum response length, and more.
  • You can expose your augmentor(s) to content creators by adding a CKEditor button, or by adding fields to your site’s entity forms. For each field use can choose the widget to use, and how it should interact with any existing data in the field you want to target.
  • For example, you could have it generate a summary from your body field and have it automatically populate the summary field. Or, you can have it suggest tags, but the specialized widget renders each suggestion as a clickable element that will add the tag to a core tag field.
  • There are currently modules available to integrate Augmentor AI with ChatGPT, Google Cloud Vision, AWS AI, and more.