The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Will ChatGPT take my job? - #608
24 snips
Dec 26, 2022 
In this intriguing conversation with ChatGPT, a groundbreaking AI language model from OpenAI, the discussion dives into the pressing question of whether AI will take our jobs. ChatGPT explores its own capabilities in conversational context, revealing both opportunities and challenges for the workforce. The model discusses how it interprets human emotions and the nuances of effective communication, while also reflecting on the irreplaceable human touch in interviews. Finally, listeners are left to ponder the future role of AI in their careers.
AI Snips
Chapters
Transcript
Episode notes
ChatGPT Refuses Impersonation
- Sam Charrington tried to make ChatGPT impersonate him, but it refused.
- ChatGPT is programmed not to impersonate specific individuals, as a safety measure.
Inconsistency in Verbal Fillers
- The guest ChatGPT's responses sometimes include verbal fillers like "ahs" and "ums" for a human-like effect.
- It's surprising how difficult it is to get ChatGPT to consistently use these fillers, despite being able to adopt other verbal affects.
Multi-Part Questions
- The interviewer ChatGPT asks multiple questions per turn, unlike Sam Charrington's usual style.
- This might be how humans naturally ask questions, learned from the training data.