# Making sense of AI as a new designer | Figma for Edu Early Career Week Source: https://www.youtube.com/watch?v=xrTBr4igMxQ Recap page: https://rapidrecap.app/video/xrTBr4igMxQ Generated: 2025-12-08T12:33:05.861+00:00 --- ## Quick Overview Product designers working on AI experiences at Figma, Duolingo, Notion, and Ramp discussed principles like maintaining user control and trust, and shared that AI accelerates workflows by acting as a thought partner, while emphasizing that human craft and taste remain crucial for refining AI-generated outputs. **Key Points:** - Panelists include Tammy (Figma AI experiences), Robert (Duolingo), Pavan (Ramp Policy Agent), and Jen (Notion AI search/connectors), all currently designing AI-driven product features. - Robert's design principles for Duolingo's AI feature, 'Video Call,' center on leveraging AI's ability to improv, always giving users the ability to take back control, and making experiences immersive without replacing the human. - Jen prioritizes principles like ensuring the AI feels like a 'partner not a black box,' giving users visibility into why AI acted, and grounding AI in real user goals by focusing on trust and predictable behavior over novelty. - Workflow acceleration is a key benefit, with panelists using AI as a thought partner to expand ideas, find edge cases, and move quickly from idea to prototyping, as Tammy noted, allowing for deeper communication. - Pavan uses AI tools like Claude to debate opinionated takes by simulating debates with other principal product designers, strengthening his position before design reviews. - Panelists agree that human craft and taste are essential to elevate AI output, which often only gets work '60% of the way there,' making the curation of context and steering of prompts the designer's unique contribution. - Advice for early career designers includes getting comfortable being a generalist, understanding foundational knowledge across disciplines, and continuously training the muscle of interacting with AI through extensive tinkering and experimentation. **Context:** This session, part of Figma's Early Career Week, featured a panel moderated by Rodrigo Davies from Figma's AI experiences team, bringing together product designers working on cutting-edge AI features at major tech companies to advise emerging designers on navigating the integration of AI into design practice. The panelists introduced themselves, sharing their educational backgrounds (ranging from Neuroscience/CS to graphic design/no formal college degree) and current roles, such as designing Duolingo's AI-powered 'Video Call' feature or Notion's AI search capabilities. ## Detailed Analysis The discussion heavily focused on establishing guiding principles for designing with AI, emphasizing user trust and control. Robert detailed his approach at Duolingo, ensuring users can interrupt the AI character Lily, whose personality is defined by specific prompts (e.g., being afraid of spiders or vegan). Jen stressed making AI a 'partner not a black box' by providing clarity on its decisions and focusing on meaningful tasks rather than novelty. The panelists universally confirmed that AI significantly accelerates workflows, transforming AI into a 'thought partner' that helps expand ideas and rapidly prototype, allowing designers to communicate concepts faster through motion rather than static sketches. For speeding up tedious tasks, Jen uses AI for copywriting generation, while Tammy noted that using AI for bulk editing or creating Figma plugins can automate drudgery. A major recurring theme was preserving human craft; the consensus is that AI provides the first 60% of output, and it is the designer's taste, intuition, and ability to curate context that refines the output to 100%. Furthermore, the panel addressed collaboration, suggesting that the lines between design, PM, and engineering are blurring, pushing designers to become comfortable generalists who understand foundational aspects of other roles, exemplified by Robert's '60/40 rule' where 40% of effort overlaps with other roles, while the sacred 60% remains specialized craft. ### Panelist Introductions and AI Focus - Tammy (Figma Prompt to Edit) - Robert (Duolingo AI speaking practice) - Pavan (Ramp Policy Agent) - Jen (Notion AI search/connectors) ### Principles for AI-Driven Design - Robert focuses on improv ability and user control - Jen emphasizes AI as a partner, avoiding black-box behavior - Trust, reversibility, and predictable behavior matter more than flashy capabilities ### AI Impact on Designer Workflows - Tammy uses AI to expand initial ideas and prototype quickly - Pavan role-plays with AI models (like Claude) to debate ideas and iterate faster - Jen uses AI for quick copywriting and learning technical details like React syntax ### Maintaining Human Craft and Taste - Robert intentionally keeps physical whiteboarding manual to keep senses sharp - Pavan prioritizes sketching his own ideas first to build opinion before relying on AI generation - Tammy defines craft as curating the context and information provided to the AI ### Shifting Team Collaboration in the AI Era - Pavan advises designers to be comfortable generalists as lines blur between design, PM, and engineering - Robert introduced the '60/40 rule,' suggesting designers maintain unique craft (60%) while handling 40% of other roles' soft skills - Jen suggests continuous experimentation with AI capabilities to train the unfamiliar muscle of AI interaction ### Advice for Early Career Designers - Pavan notes the leveled playing field allows for experimentation but urges focusing this on solving user needs - Robert warns against using AI as an executioner rather than a collaborator - Tammy encourages designers to tinker, ask 'why' about AI outputs, and pay attention to prompting styles.