Maai & AI

Tagged: conversation-design

61 entries

🌳 videos

Designing for doubt

Before 2022, it was pretty clear whether you were talking to a bot or a human. With the arrival of ChatGPT, this changed – radically. With a user interface that combines the gift of the gab with information that's not necessarily accurate, designers are faced with a novel challenge. How do we design for interfaces that are so convincing that people instinctively drop their guards and trust them more than might be good for them? And do our traditional design paradigms still serve us here?
🌳 videos

Content driven interaction in 4 steps

A practical guide to designing better AI interactions with documents using a systematic four-step approach that prioritizes content analysis, user needs, and task design. Content-driven interaction moves beyond simple Q&A patterns to create dynamic, purpose-driven AI interactions that leverage the intentional design embedded in documents. Analyzing your source content is the critical first step often overlooked: understanding what is in your documents, how they are structured, and what information is relevant prevents hallucinations and improves AI accuracy. The four-step methodology (analyze content, analyze users and tasks, design role and task, design interaction) creates a systematic approach to building RAG systems that actually serve user needs. Practical implementation includes welcome messages, rephrase-and-expand techniques, chain-of-thought reasoning, and step-by-step workflows that guide users through complex information.
🌳 videos

Kudos to Chevrolet!

An analysis of Chevrolet's GPT-based chatbot vulnerability and how the company successfully addressed the security issues through proper monitoring and design improvements. The initial vulnerability was not a sign of stupidity, but a necessary part of the testing process that all chatbots must undergo. The real measure of competence is not whether vulnerabilities exist, but how quickly and effectively they respond. I test the updated bot and demonstrate practical prompt engineering techniques for controlling chatbot behavior, including defensive prompting, topic maintenance, and firm responses to abuse. GPT-based solutions alone are not sufficient for high-risk transactions: a hybrid approach combining multiple design techniques is more effective.
🌳 videos

Bard violates Grice's maxims...again

A critical analysis of Google Bard failing to adhere to conversational principles by withholding relevant information about Gemini model availability in the EU. Bard violates Grice's Cooperative Principle by failing to proactively communicate that Gemini is unavailable in the EU due to regulatory reviews. The issue is not a hallucination but a failure of the maxim of quality: Bard does not provide necessary information unless explicitly asked. Users must already know the answer to their question before asking it, creating a paradoxical and unhelpful interaction pattern. Bard in Europe still runs on the older Lambda/PaLM model, not the newer Gemini model available in other regions. This violation of conversational principles is arguably worse than generating false information because it presupposes user knowledge.
🌳 videos

Prompt tutorials for conversation designers

A tutorial on three essential prompting techniques to make ChatGPT more conversational and natural. I demonstrate the problem with ChatGPT's default responses, which are too long, impersonal, and unengaging for conversational interfaces. The three tips: use OpenAI Playground instead of ChatGPT for better control over LLM behavior and access to system prompts, define a specific expert role in the system prompt rather than the generic helpful assistant since research shows expert roles result in higher quality answers, and prompt for specific behaviors like conciseness, staying on topic, and ending messages with questions to drive conversation forward. I use structured delimiters to organize prompt sections and demonstrate how each behavioral instruction refines the bot's personality and interaction style. The process is like chipping away at a block of marble: each prompt chips away at behavior, domain, interaction models, and persona until you get the conversational bot you want.
🌳 videos

Van technisch schrijver naar conversation designer

A presentation exploring my transition from technical writing to conversation design, discussing how communication professionals are evolving their roles in the digital landscape. I explain that conversation design goes beyond simple chatbot scripting or dialogue trees, involving understanding user needs, communication patterns, and designing interactions that feel natural and helpful. Technical writers are well-positioned for this transition because their core skills, clarity, user focus, and structured thinking, are exactly what conversation designers need. The shift is not about learning entirely new skills but about applying existing expertise in a new context, leveraging experience with documentation, user guides, and information architecture while expanding into dialogue design, tone and voice development, and interaction patterns. The future of technical communication is conversational, and the intersection of technical writing and conversation design represents how organizations will interact with their users in an increasingly digital world.