The Automotive Leaders Podcast · Jan Griffiths

AI, Trust, and the Human Shift: What Automotive Leaders Must Do Next

·36 min·2 clips

The question is, are we delegating to the machine or are we collaborating with the machine?

1. The Automotive Leaders Podcast, hosted by Jan Griffiths, follows up an earlier episode with a discussion of how automotive leaders can move from casual ChatGPT use to practical AI adoption, including trust, culture and agentic AI. 2. Griffiths, who describes herself as a farmer's daughter from Wales with over 35 years in the auto industry, is joined by Dr. Brian O'Brien, an MIT research scientist and author of 'How to Make AI Useful', and Cheryl Thompson, CEO of Cadia, who has taken weekly AI classes since April. 3. The central thesis from O'Brien is that large language models are 'a phenomenal Copilot' but 'not an autopilot', so leaders should start small, scale smart and keep humans in the loop. 4. Thompson explains the 'wow, whoa, grow' framework from the previous episode: after the 'wow' of building with AI, she found herself creating more work instead of freeing time, and now meditates to keep her 'leading AI' rather than the reverse. 5. Griffiths and Thompson give examples of AI errors, including a Claude workshop assistant that fabricated data despite a 'do not fabricate' instruction, and a custom GPT that ignored a 1,300-character LinkedIn limit and then replied that it was 'focusing on those three other things'. 6. O'Brien says models fail unpredictably and so will not solve safety-critical problems, and he predicts that large language models are 'a fad' that will give way to whatever comes next, as deep neural networks did. 7. Thompson describes use cases such as prompting a workshop assistant to respond as author Henry Cloud, converting technical documents into analogies, running a 14-day calendar review through 365 Copilot every Friday at five, and timing Thanksgiving biscuits during a Lions game. 8. The group cites a Women in Manufacturing Conference slide showing that Gen X trusts AI most and boomers least, while Griffiths' college-student daughter refuses to use it because she thinks it will make people stupid; O'Brien responds that lower neural activity in studies reflects delegation rather than collaboration. 9. On agentic AI, Thompson describes a procurement future with 'an army of agents' making sourcing decisions, and O'Brien argues that human negotiators built on 20-year relationships may outperform bots because of the 'emotional component'. 10. Leaders in automotive manufacturing and supply who want practical steps beyond basic ChatGPT use will find this useful.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from The Automotive Leaders Podcast