
🎙 EP 368: Trump War Grok Consultation & "Mode-Hopping" Pre-Training Discovery
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!
Reports revealed President Trump consulted xAI's Grok chatbot regarding political dynamics in Venezuela ahead of strategic military operations, highlighting AI's growing footprint in executive decision-making. Meanwhile, a machine learning research paper uncovered "mode-hopping" during LLM pre-training, proving that additional training tokens can cause drastic, non-linear collapses in downstream reasoning and fine-tuning transferability.
We’ll talk about:
Details outlining executive use of…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.