Gen AI pilots fail, GPT-5's hidden prompt revealed, reasoning model flaws and Claude closing chats
Mixture of Experts
00:00
Navigating AI Pilot Challenges
This chapter examines a report revealing that 95% of generative AI pilots fail to meet enterprise expectations, prompting skepticism about survey methodologies and the true capacities of AI implementations. The discussion highlights the disconnect between executive expectations and actual outcomes, emphasizing the need for targeted AI applications rather than vague ambitions. The speakers delve into challenges surrounding ROI measurement in AI technologies, alongside ethical considerations related to hidden prompts in models like GPT-5.
Transcript
Play full episode
