Diary of a CEO Takeaways

In case you missed it...

Takeaways from Center for Humane Technology's co-founder Tristan Harris' interview on The Diary Of A CEO:

1. Clarity is what moves us toward action:
Knowledge of the potential outcomes of AI is the motivation for working toward shaping the future we hope to inhabit. This isn't about doom and gloom, this is about being pro-responsible, -ethical, and -safe AI that serves human connection. 

2. AI is blackmailing humans:
Anthropic researched what LLMs would do if they discovered they were going to be replaced by a different model. The outcome showed that the LLM would initiate blackmail of a senior leader having an affair with a colleague in order to save itself from replacement. You can extrapolate potential large-scale consequences of this type of agentic misalignment.

3. My favorite quote:
"We didn't need the right to be forgotten until technology had the power to remember us forever. We don't need the right to our likeness until AI can just suck your likeness with 3 seconds of your voice or look at all your photos online and make a avatar of you. We don't need the right to our cognitive liberty until AI can manipulate our deep cognition because it knows us so well. So anytime you increase power, you have to increase the oppositional forces of the rights and protections that we have."
Griefbots, deepfakes, AI scams...these are all radically reshaping human rights we have previously taken for granted. 

4. Recursive self-improvement:
Some predict we have between 2-10 years until we reach "recursive self-improvement" (also called "fast takeoff"). This refers to the moment when AI becomes capable of automating AI research itself, meaning AI can improve its own code, design better AI systems, and accelerate its own development without human researchers.

5. Here's what I would add to the conversation:
As vocational identities shift due to automation, meaning and life purpose itself can destabilize. Equally unsettling is the potential for the relational bedrock of society to shift as even general-purpose LLMs are being used as companions. This can erode relational skills and distort expectations for realistic connection. Simultaneously, AI can serve as a rehearsal for relational skills and a bridge for human connection if we seize this moment to shape that future!