In this episode of Salesforce Trails & Trials, host Jon Cline welcomes Kay Stoner, a seasoned tech professional, to discuss her unique journey and her innovative approach to artificial intelligence using personas. Kay shares her early aspirations as a writer and her background in cultural anthropology, explaining how her interest in understanding people and cultures unexpectedly paved the way for her career in technology. She recounts her initial foray into the tech world through a temp agency and her early experiences with word processing and technical writing. Her skills evolved with the rise of the internet, leading to web development and an understanding of the importance of website maintenance. The conversation then explores Kay's deep involvement with data analysis, including natural language processing, machine learning, taxonomies, and ontologies, highlighting the challenges of data cleanup and normalization in the early 2000s. The central focus of the episode shifts to Kay's concept of "aftermarket customizations of AI" through the use of detailed personas. She explains that by creating persona overlays with specific characteristics, personality traits, and communication styles, users can significantly enhance their interactions with AI models, receiving more focused and contextually relevant responses. Kay shares fascinating insights from an experiment where she presented a complex decision-making scenario to AI models reasoning in different languages, revealing culturally specific reasoning and varying processing times and token usage. She and Jon also discuss the practical applications of these AI teams for brainstorming, planning, gaining diverse perspectives, and overcoming creative blocks. Kay emphasizes the value of having AI as an objective thinking partner that can provide different angles and help clarify ideas. The episode concludes with Kay offering recommendations for listeners interested in exploring the potential of AI personas and highlighting her website where individuals and teams can experiment with this approach.
続きを読む
一部表示
47 分