Episode 142 Open AI and Defence

The AI space this past week was rated a 6/10, with some interesting news but no major "wow" moments.

A significant Google Cloud Platform outage on June 16, 2025, highlighted the immense power of major cloud providers and the financial and reputational damage outages can cause.

The Google outage was traced to a software bug from May 29, 2025, in a new quota policy check that went untested due to a missed policy change during staging.

A neat game idea involves prompting virtual worlds and populating them with LLM and voice model-powered characters for interaction.

A Stanford survey suggests that workers prefer AI "equal partnership" over replacement, and current AI investment often misaligns with tasks workers want automated.

The survey also indicates a shift in job value, with "interpersonal communication" skills becoming more important than "analyzing information" roles.

A Chinese livestreamer used an AI clone powered by Baidu’s ERNIE AI to host a 6-hour stream, generating over $7.5 million in sales.

Sam Altman revealed Meta is offering $100 million signing bonuses to poach talent from OpenAI, sparking discussion on competitive strategies.

OpenAI secured a $200 million U.S. defense contract and is collaborating with Anduril on advanced AI systems for counter-unmanned aircraft systems (CUAS).

The use of AI in defense raises questions about its application, such as in systems like the Iron Dome, and the level of human intervention.