Davidmanheim

A Per­sonal (In­terim) COVID-19 Postmortem

Mar­ket-shap­ing ap­proaches to ac­cel­er­ate COVID-19 re­sponse: a role for op­tion-based guaran­tees?

Po­ten­tial High-Lev­er­age and In­ex­pen­sive Miti­ga­tions (which are still fea­si­ble) for Pandemics

Ineffec­tive Re­sponse to COVID-19 and Risk Compensation

Link: Does the fol­low­ing seem like a rea­son­able brief sum­mary of the key dis­agree­ments re­gard­ing AI risk?

Up­dat­ing a Com­plex Men­tal Model—An Ap­plied Elec­tion Odds Example

Theater Tick­ets, Sleep­ing Pills, and the Idiosyn­crasies of Del­e­gated Risk Management

Diver­gence on Ev­i­dence Due to Differ­ing Pri­ors—A Poli­ti­cal Case Study

Hack­able Re­wards as a Safety Valve?

[Question] What Pro­gram­ming Lan­guage Char­ac­ter­is­tics Would Allow Prov­ably Safe AI?

Mesa-Op­ti­miz­ers and Over-op­ti­miza­tion Failure (Op­ti­miz­ing and Good­hart Effects, Clar­ify­ing Thoughts—Part 4)

Ap­ply­ing Overop­ti­miza­tion to Selec­tion vs. Con­trol (Op­ti­miz­ing and Good­hart Effects—Clar­ify­ing Thoughts, Part 3)

What does Op­ti­miza­tion Mean, Again? (Op­ti­miz­ing and Good­hart Effects—Clar­ify­ing Thoughts, Part 2)

Re-in­tro­duc­ing Selec­tion vs Con­trol for Op­ti­miza­tion (Op­ti­miz­ing and Good­hart Effects—Clar­ify­ing Thoughts, Part 1)

Schel­ling Fences ver­sus Marginal Thinking

Values Weren’t Com­plex, Once.

Over­sight of Un­safe Sys­tems via Dy­namic Safety En­velopes

Col­lab­o­ra­tion-by-De­sign ver­sus Emer­gent Collaboration

Multi-Agent Overop­ti­miza­tion, and Embed­ded Agent World Models

Policy Beats Morality