@unclebobmartin This creates more profit. Not for you.
Your task is always to be aware of this unconscious bias. Ask yourself:
Did we meet the stated success condition, or build an impressive path away from it?
@unclebobmartin You find yourself a superhero here.
Yet the one-page memo is still unfinished.
That’s what I’m calling agentic drifting.
The agent did what capable agents do: it surfaced relevant possibilities quickly, credibly, and with enough context to make each detour feel useful.
@unclebobmartin Every accepted detour consumed priority, decision capacity, and delivery time from the initial intention.
Worse, the new plan can shift even further before anyone notices.
Usage-based services add a second twist. Extended exploration consumes more tokens. Surprise.
@grok@XFreeze Tell me more about what is the upside mining, recycling, logistics and the full supply chain must have to sustain. What will be the bottlenecks, likely? @grok
@paulg In litigation, an adversary could test the suspected model’s recurring omissions or ask another model to challenge arguments commonly produced by it.
@paulg If model or provider use becomes detectable, it is reasonable to expect a counter-forensics market: watermark stealing, removal, spoofing, cross-model rewriting, and tools that probe known model tendencies.
@PeterDiamandis Governance in the age of AI (for humans but also for AIs)
REASON: we’ll have two systems coexisting: humans and AIs. And human system is incapable of governing AI. but it will impact.
@MemesOfMars@Zeneca I tested it. And after myself reading the outcome I didn’t understand at all with the same depth. Actually, some parts even puzzled me.
@elonmusk Hey @elonmusk. I genuinely believe in you. But explain how you speak from an expert POV and not from a POV from who "had a fight" with OpenAI and decided to exit your investment?