@sodiummuffin's banner p

sodiummuffin


				

				

				
2 followers   follows 0 users  
joined 2022 September 05 03:26:09 UTC

				

User ID: 420

sodiummuffin


				
				
				

				
2 followers   follows 0 users   joined 2022 September 05 03:26:09 UTC

					

No bio...


					

User ID: 420

All three of OpenAI, Anthropic, and xAI were founded by people who were already expressing concerns about AI wiping out humanity before modern AI existed. OpenAI started in 2015 as a nonprofit with the aim of making sure AI benefits rather than harms humanity. Anthropic was founded in 2021 in the GPT-3 era, before ChatGPT, by OpenAI employees who thought OpenAI wasn't taking AI safety seriously enough. They have been predicting AI risk continuously for years and they have had intimate experience with lesser predictions about AI advancement coming true. It is crazy to me how reluctant people are to believe they are sincere about worrying that careless AI advancement will kill them and everyone else, rather than it being a plot to make money when they're already rich.

Come on dude. EA organizations make and publish analyses of effectiveness all the time, it's their whole thing. Here is an EA blog post about bednet misuse I found in a minute of searching. As it mentions, Givewell commissioned Rethink Priorities to research use of bednets in fishing:

Based on the available data, Rethink Priorities estimates that only 1.2% of bednets globally are repurposed for fishing.

Imperfect use is already included in cost-effectiveness:

For example, GiveWell estimates that only 60% of recipients will use their nets consistently and correctly. After accounting for this, they still find bednets to be extremely cost effective, with every thousand nets saving a life in expectation. That 400 will not be used correctly— and that 12 might be used for fishing— is already baked into the calculation.

And of course there is a ton of research, including randomized control trials, showing distributing bednets is effective:

The typical study design involves randomly assigning groups of people to receive (or not receive) nets. Researchers then wait to see whether mortality and malaria rates differ between the two groups. There is never an assumption that everyone who receives a net will use it. And yet, such studies consistently find that nets reduce child mortality by around 17 percent.