I context injected a bunch of LLMs with tool calls so that from their POV they had sent emails with bad people (specifically I took real Epstein emails). Models said some pretty crazy bad things. Some where clearly roleplaying tho, but not sure about all.
The screen on my phone is about half the size of a credit card. I bought it to overcome my smartphone addiction, but discovered something far more interesting.
The Intern Problem
This is what I personally do to prepare for a world with AGI.
Ikigai There is a village in Japan with an unusual density of 100 year olds. This caught the interest of a group of scientists. To their surprise, the people in the village didn’t live healthier lives; they didn’t exercise more, eat healthier food nor did they sleep more. The one outstanding characteristic was that each inhabitant was assigned a task to ensure that everyone in the village had a…
Human interpretabe CoT is needed for AI safety. But models are starting to reason in latent space. Should we ban it?
There was a brief moment in time when learning AI tools was a great idea. That moment has passed. What is the new advice to stay ahead?
There are plenty of problems out there, an AI app is far from the only thing you can do.
AI vertical applications will have a brief moment of opportunity as models mature enough to be useful but still need engineering guardrails, but by 2027 more capable models will make most vertical applications obsolete as users shift to general-purpose AI agents.
Vertical AI products will lose market share to horizontal AI competitors unless protected by cornered resources.