the fires in LA are currently burning through about 5 football fields a minute. over 2,900 acres have been burned and over 80,000 people have had to evacuate. so many who lost homes. such a devastating and frightening situation.
A key feature of copaganda is that police and the news media attempt to use crises to increase the size, power, and profit of the punishment and surveillance bureaucracies. This has long been one of the creepiest things about it. They don’t let a good crisis go to waste.
Ok, here's something I learned in high school humanities: As long as people believe that they have redress for injustice, either through courts or through political action (like voting), the people will choose nonviolent options.
🧵
Entropy is one of those formulas that many of us learn, swallow whole, and even use regularly without really understanding.
(E.g., where does that “log” come from? Are there other possible formulas?)
Yet there's an intuitive & almost inevitable way to arrive at this expression.
my intention was to find a way to condition him that nothing on screens is particularly interesting or engaging
i spent time thinking how to do this and remembered reading about public access channels in norway
i decided to run an experiment with slow tv
"Large language models surpass human experts in predicting neuroscience results" w @ken_lxl and https://t.co/YOhCmQlJsu. LLMs integrate a noisy yet interrelated scientific literature to forecast outcomes. https://t.co/49WYirBdBv 1/8
This is a great paper. It points out:
1. Humans do not even approximately behave according to rational choice theory
2. There is no reason to think advanced AI will "inevitably" maximize some utility function
3a. Human preferences are derivative / constructed, so aligning AI by matching its behavior to our stated preferences is wrongheaded
3b. We can align AIs directly to some normative ideal of a "good assistant / programmer / driver / etc." instead
4. Aggregating preferences across people is fraught with philosophical and mathematical difficulties. We should not aim to align AI to the "collective will of humanity."
@Boenau@Waymo I live in SF, where Waymos are everywhere, and when I’m at a pedestrian cross would prefer the oncoming traffic to be Waymos 100% of the time. They come to a full stop well in advance of the crosswalk. Human drivers regularly blow past me while I’m in the middle of the street.
I will be joining Patrick House (@patrick__house) on stage at the Warwick Arts Centre in Coventry on 23rd October. Patrick, a neuroscientist and writer, is known for his work on free will and consciousness. After our discussion, there will be a Q&A session. I look forward to what promises to be an engaging evening, and I hope to see you there.
@strubell@JeffDean Indirectly related and critical: it would be great if Google disclosed the carbon footprint of their AI summary in searches and allowed users to opt out. I don't want my simple searches to have additional, unnecessary carbon emissions for a capability I don't need
Great thread. In a similar vein, I'm gonna take this opportunity to take this hot, controversial & incredibly politicised topic to make a couple of points about measurement theory (which usually makes people go to sleep). 1/
One reason I'm excited about the Llama 3 paper is we finally have a description of what you need to do to get an LLM to reason sort-of well (human supervision, symbolic methods in post-training). GPT-4 and co. do not show that reasoning "emerges from distributional learning".
Fellow researchers, just had a series of eye-opening emails with our grants/contracts/legal ppl about applying for OpenAI's Superalignment grant https://t.co/nsm8TTB4Q5 It appears @OpenAI 's terms are not aligned with the kind of contract public institutions can enter into.