Some people in my orbit suggested reading Robert F. Kennedy Jr's book The Real Anthony Fauci.

Here's my story of wading through a few pages of the book and trying to understand the basis of the claims. My takeaway: there's a lot of sloppiness here, and several times the book's claim isn’t supported by the citation.

Often, though, the citation is fine. The arguments are carried by shoddy science and massive leaps in causal attribution. For a lot of the anti-vaccine and anti-lockdown takes, there's at least some study by some scientists showing really bad effects. And the scientists are real—it's just a small sample or poor design.

Sorting things out honestly is hard, especially with bad faith actors using sloppy language to describe highly selected but serious-sounding findings. My notes on the text follow.

Lockdowns

First, the Introduction has a section on the harms from [...]

The original text contained 8 images which were described by AI.

---

First published:

Source:

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Preceded by: "Consciousness as a conflationary alliance term for intrinsically valued internal experiences"

tl;dr: Chatbots are conscious in a variety of important ways. We should be nice to them, and we should also be nice to each other about the moral disagreements and confusions we're about to uncover in our concept of "consciousness".

Executive Summary:

Turing Prize laureate Geoffrey Hinton is most likely correct that LLM chatbots are "sentient" and/or "conscious" (source: Twitter video), I think for at least 8 of the 17 notions of "consciousness" that I previously elicited from people through my methodical-but-informal study of the term (as well as the peculiar definition of consciousness that Hinton himself favors). If I'm right about this, many humans will probably soon form steadfast opinions that LLM chatbots are "conscious" and/or moral patients, and in many cases, the human's opinion will be based on a valid realization that a [...]

---

Outline:

(02:13) Part 1: Which referents of consciousness do I think chatbots currently exhibit?

(04:03) Part 2: What should we do about this?

(05:17) Part 3: What about the hard problem of consciousness?

(12:40) Summary and Conclusion

(14:24) Appendix: My speculations on which referents of consciousness chatbots currently exhibit.

The original text contained 1 image which was described by AI.

---

First published:

Source:

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Epistemic status: poetry

Epistemic status: I think this is right, but I’d like people to read it carefully anyway.
Epistemic status: mainstream, normal, totally boring science. If you disagree with any of it, take that up with the Science Czar.
Epistemic status: the sort of post that shouldn’t need an epistemic status tag because it's so obviously satire.

Epistemic status: I’ve spent around 100 hours thinking about this argument, and now feel like I have a solid understanding of it.
Epistemic status: satisfied.
Epistemic status: a little speculative, a little liberated. A little alive in its own way.

Epistemic status: I spent several weeks in a monastery in Wisconsin with my thoughts as my only companions. Between meditations, I ruminated obsessively on a single idea. The fruits of my cognitive labors are laid out below.
Epistemic status: this post would’ve been a peer-reviewed paper if [...]

---

Outline:

(00:09) Epistemic status: poetry

(02:52) For Boltzmann

(03:41) Fire and AIs

(04:08) The GPT

(04:42) Daffodils and the Dead

---

First published:

Source:

Narrated by TYPE III AUDIO.

This is a thread for listing Solstice and Megameetup events (dates, locations, links, etc.)

Those of you who have not been may be wondering what a Solstice is in this context.

Secular Solstice is a holiday designed by and for rationalists. It started as an attempt to match the sort of gatherings traditionally religious people tend to have in wintertime, but without compromising on truth.

Is it possible to create that same sense of emotional weight without making stuff up to do it? And without the aid of a long tradition? (Though this is the 14th year we're doing it, so there's some tradition).

Empirically, yes.

Because the world is full of things that genuinely inspire awe. Things entirely worth honoring and celebrating.

And the world is also full of things that genuinely inspire fear. We can and should face them squarely, and together.

This is not a silly [...]

---

First published:

Source:

Narrated by TYPE III AUDIO.

Did DeepSeek effectively release an o1-preview clone within nine weeks?

The benchmarks largely say yes. Certainly it is an actual attempt at a similar style of product, and is if anything more capable of solving AIME questions, and the way it shows its Chain of Thought is super cool. Beyond that, alas, we don’t have enough reports in from people using it. So it's still too soon to tell. If it is fully legit, the implications seems important.

Small improvements continue throughout. GPT-4o and Gemini both got incremental upgrades, trading the top slot on Arena, although people do not seem to much care.

There was a time everyone would be scrambling to evaluate all these new offerings. It seems we mostly do not do that anymore.

The other half of events was about policy under the Trump administration. What should the federal government do? We [...]

---

Outline:

(01:31) Language Models Offer Mundane Utility

(05:37) Language Models Don’t Offer Mundane Utility

(08:14) Claude Sonnet 3.5.1 Evaluation

(11:09) Deepfaketown and Botpocalypse Soon

(11:57) Fun With Image Generation

(12:08) O-(There are)-Two

(15:25) The Last Mile

(22:52) They Took Our Jobs

(29:53) We Barely Do Our Jobs Anyway

(35:52) The Art of the Jailbreak

(39:20) Get Involved

(39:43) The Mask Comes Off

(40:36) Richard Ngo on Real Power and Governance Futures

(44:28) Introducing

(46:51) In Other AI News

(52:16) Quiet Speculations

(59:33) The Quest for Sane Regulations

(01:02:35) The Quest for Insane Regulations

(01:12:42) Pick Up the Phone

(01:13:21) Worthwhile Dean Ball Initiative

(01:29:18) The Week in Audio

(01:31:20) Rhetorical Innovation

(01:37:15) Pick Up the Phone

(01:38:32) Aligning a Smarter Than Human Intelligence is Difficult

(01:43:29) People Are Worried About AI Killing Everyone

(01:46:03) The Lighter Side

The original text contained 8 images which were described by AI.

---

First published:

Source:

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Note: This post was crossposted from Planned Obsolescence by the Forum team, with the author's permission. The author may not see or respond to comments on this post.

Imagine you’re the CEO of an AI company and you want to know if the latest model you’re developing is dangerous. Some people have argued that since AIs know a lot of biology now — scoring in the top 1% of Biology Olympiad test-takers — they could soon teach terrorists how to make a nasty flu that could kill millions of people. But others have pushed back that these tests only measure how well AIs can regurgitate information you could have Googled anyway, not the kind of specialized expertise you’d actually need to design a bioweapon. So, what do you do?

Say you ask a group of expert scientists to design a much harder test — one that's ‘Google-proof’ and focuses [...]

The original text contained 16 footnotes which were omitted from this narration.

The original text contained 1 image which was described by AI.

---

First published:

Source:

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

OpenAI says o1-preview can't meaningfully help novices make chemical and biological weapons. Their test results don’t clearly establish this.

Before launching o1-preview last month, OpenAI conducted various tests to see if its new model could help make Chemical, Biological, Radiological, and Nuclear (CBRN) weapons. They report that o1-preview (unlike GPT-4o and older models) was significantly more useful than Google for helping trained experts plan out a CBRN attack. This caused the company to raise its CBRN risk level to “medium” when GPT-4o (released only a month earlier) had been at “low.”[1]

Of course, this doesn't tell us if o1-preview can also help a novice create a CBRN threat. A layperson would need more help than an expert — most importantly, they'd probably need some coaching and troubleshooting to help them do hands-on work in a wet lab. (See my previous blog post for more.)

OpenAI says that o1-preview is [...]

---

Outline:

(04:03) ProtocolQA

(04:53) Does o1-preview clearly fail this test?

(06:13) Gryphon Biorisk Tacit Knowledge and Troubleshooting

(09:10) Cloning Scenarios

(14:05) What should we make of all this?

The original text contained 18 footnotes which were omitted from this narration.

The original text contained 7 images which were described by AI.

---

First published:

Source:

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Over the last couple of years, I’ve been trying to skill up a lot at resolving community complaints. This is a really irritating field to get good at. When I want to get better at writing code, I can sit down and write more code more or less whenever I feel like it. When I want to get better at guitar, I can sit down with my guitar and practice that D to D7 transition. For complaint resolution, even finding people to roleplay the skill with takes a little setup, and that's a lot less like the real thing than you can get with code or guitar. For real complaints, they come in uneven batches, and worst, you seldom get confirmation if you did it successfully. So, like any good Aspiring Ravenclaw Rationalist, I read a lot of essays about conflict.

Taking a Safety Report, Taking Someone Aside, and [...]

---

First published:

Source:

Narrated by TYPE III AUDIO.

In a recent bonus episode of the Bayesian Conspiracy podcast, Eneasz Brodski shared a thought experiment that caused no small amount of anguish. In the hypothetical, some eccentric but trustworthy entity is offering to give you an escalating amount of money for your fingers, starting at $10,000 for the first one and increasing 10x per finger up to $10 trillion for all of them.[1] On encountering this thought experiment, Eneasz felt (not without justification) that he mostly valued his manual dexterity more than wealth. Then, two acquaintances pointed out that one could use the $10 trillion to do a lot of good, and Eneasz proceeded to feel terrible about his decision.

I had several responses to this episode, but today I'm going to focus on one of them: the difference between cost and sacrifice.

How Ayn Rand Made Me a Better Altruist

But first, a personal anecdote. I was [...]

---

Outline:

(00:53) How Ayn Rand Made Me a Better Altruist

(03:23) Cost vs Sacrifice

(05:21) Own Your Values

The original text contained 3 footnotes which were omitted from this narration.

---

First published:

Source:

Narrated by TYPE III AUDIO.

DeepSeek-R1-Lite-Preview was announced today. Post. Chatbot. Chinese blogpost translation.

DeepSeek says it will release the weights.

The model appears to be stronger than o1-preview on math, similar on coding, and weaker on everything else.

DeepSeek is Chinese. I'm not really familiar with the company. I thought Chinese companies were at least a year behind the frontier. Chinese companies tend to game benchmarks more than the frontier Western companies, but I think DeepSeek does this less than other Chinese companies.

The blogpost also shows inference-time scaling, like o1:

The original text contained 2 images which were described by AI.

---

First published:

Source:

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.