LessWrong posts by zvi

Introducing Plan A

The folks who brought you AI 2027, a so far remarkably accurate set of predictions despite those predictions having seemed freaky to many at the time, now bring you their positive vision that involves more freaky predictions: Plan A.

These guys have rather strong prediction track records. In addition to AI 2027, among other things, Daniel Kokotajlo has What 2026 Looks Like (which is remarkably similar to what 2026 looks like) and Ryan Greenblatt, who is also the chief scientist at Redwood Research, was the #2 most accurate AI forecaster in 2025 out of 413 entries. Past performance is as always no guarantee of future success.

If you’re the type to read at least some of my posts, or if you thought AI 2027 was worth reading, I recommend reading Plan A.

There is also an unofficial visual novel version, for minds very different from my own who would want that.

To be clear up front: I am not endorsing Plan A. I am not suggesting we should go off and try to enact Plan A as written. There is a lot more work to do and a lot of potential [...]
---
Outline:
(00:09) Introducing Plan A
(01:42) You Only Get Five Words
(08:44) Proactive Response To Objections
(09:30) Initial Introductions and Endorsements
(14:50) A Positive Vision
(17:53) Alternative Plans
(19:31) Plan S for Shutdown
(21:35) Something (Unexpectedly Good) Ever Happens
(23:10) Thus Selective Optimism
(24:17) Quickly, There's No Time
(25:20) Race Conditions
(27:23) This Is A Lot Of Diffusion And Economic Growth
(28:36) Living In China
(30:35) Planning For Shifting Overton Windows Is Essential
(31:51) The Standard Handwave
(35:23) Some Equate Any Controls Over Compute To Authoritarian Dystopia And Those Same People Mostly Think Superintelligence Won't Happen
(37:07) Vitalik Buterin Is Right, The Crux Is Future AI Capability Levels
(44:45) The Authoritarian Objection
(50:17) Concepts Of A Plan
(52:03) The Kitchen Sink
(53:13) Selective Claims Of Authoritarianism
(55:12) You Either Can Steer The Future Or You Cannot
(57:40) Cooperative Alignment
---

First published:

July 11th, 2026


Source:

https://www.lesswrong.com/posts/z9tXCGogEgkgHSh8G/introduction-for-and-reactions-to-plan-a

---

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.

This is part 2 of the weekly, broadly covering speculation, rhetoric and policy, along with alignment research.

This does not cover the release of GPT-5.6-Sol. As always, I will be taking a few days to digest what the new model has to offer and to allow others to try it and react. I will cover Sol and its capabilities early next week. I covered the GPT-5.6 system card back on June 28.

This also does not cover the release of Plan A, the follow-up to AI 2027. This new scenario is a positive vision of what its authors think we should do going forwards.

I do not endorse all of the recommendations or predictions of Plan A, but I do endorse reading Plan A and taking it seriously. Scott Alexander, one of those who worked on it, writes an introduction and justification here. I will have full coverage soon.

Table of Contents


Quiet Speculations. Will our AI regulations be ad hoc indefinitely?

The Goalposts Are Dyson Spheres. This might take a little longer.

People Just Say Things.

Three Pills. Unpilled, AI, AGI, ASI.

The Quest for Sane Regulations. You’ve [...]
---
Outline:
(01:08) Quiet Speculations
(04:41) The Goalposts Are Dyson Spheres
(08:11) People Just Say Things
(08:40) Three Pills
(14:52) The Quest for Sane Regulations
(16:35) OpenAI National Security Principles
(23:45) Greetings From The Department Of War
(25:47) Chip City
(27:02) Open Weight Models Are Unsafe And Nothing Can Fix This
(34:37) Their AI Propaganda Bots
(36:39) Rhetorical Innovation
(38:23) You Learn
(40:00) You May Be Tan And Thin And Rich But You're a Tool
(44:21) Train Those Thoughts
(45:16) Train Out Those Thoughts
(50:02) My Own Private Idaho
(51:46) Aligning a Smarter Than Human Intelligence is Difficult
(57:04) No Space Like J-Space
(01:06:50) Cooperative Alignments
(01:07:15) The Lighter Side
---

First published:

July 10th, 2026


Source:

https://www.lesswrong.com/posts/7Av7wwErwNXdM3J68/ai-176-part-2-plan-b

---

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.

Enough things added up that this week is getting split into two parts.

Then on Monday, if all goes as I expect, we’ll cover OpenAI's Sol, aka GPT-5.6.

OpenAI also gave us an upgraded voice mode, which I haven’t tried out but early reports are that it is a step change.

AI writing, especially Claude writing, is becoming more prominent and harder not to notice, and increasingly a tough read when encountered in the wild. Does anyone care? Or are those who care the weird ones here?

This week saw an excellent paper, which I cover in No Space Like J-Space.

Technically we also got Grok 4.5.

Table of Contents


Language Models Offer Mundane Utility. A whole new world.

Language Models Gain Unexpected Affordances. Wait, you can just do that?

Language Models Don’t Offer Mundane Utility. Things get old.

Pay The Man His Money. You have a few more days with marginally free Fable.

Huh, Upgrades. Anthropic raises API platform limits.

Grok 4.5 Exists. It might be okay for its price.

F*** It We’re Doing It Live. OpenAI gives us a big upgrade to voice [...]
---
Outline:
(00:54) Language Models Offer Mundane Utility
(03:21) Language Models Gain Unexpected Affordances
(06:12) Language Models Don't Offer Mundane Utility
(06:54) Pay The Man His Money
(07:37) Huh, Upgrades
(07:45) Grok 4.5 Exists
(09:12) F\*\*\* It We're Doing It Live
(10:24) On Your Marks
(11:13) Better Call Sol
(22:40) Get My Agent On The Line
(23:29) Deepfaketown and Botpocalypse Soon
(25:33) Fool Me Twice
(27:23) I Like Your Style
(33:28) Enough With That Style
(36:58) Fun With Media Generation
(38:51) Copyright Confrontation
(39:30) Cyber Lack of Security
(39:52) A Young Lady's Illustrated Primer
(45:56) They Took Our Jobs
(51:07) Get Involved
(52:10) In Other AI News
(56:44) Show Me the Money
(57:09) Bubble, Bubble, Toil and Trouble
---

First published:

July 9th, 2026


Source:

https://www.lesswrong.com/posts/M9eLyMsH5DLjMYL86/ai-176-part-1-doing-it-live

---

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.

We have a respite, so I thought I’d tackle various thoughts on children, phones and screens. GPT-5.6-Sol drops tomorrow, and the Fable agents are hard at work.

I’ll start with the other screens, then finish with the phones.

Table of Contents


EdTech.

NonEdTech.

Do Not Ban Social Media Outright.

Some Modern Kids Media Is Pretty Great.

Ban Phones In Schools (1).

Your Offer Is Acceptable.

Ban Phones In Schools (2).

Screen Time.

Inappropriate Content.

EdTech

Increasingly, when you pick a school, you are picking EdTech. The school will put your child on a tablet or computer, and expect them to learn that way.

In theory, with sufficient assistance and bespoke design and incentive structures, this is The Way. It sure seems way better than ‘sit and listen to a lecture.’

I am especially excited for Alpha School's version of this, with its bespoke designs and high level of both expectations and continuous human support.

Alas, most people are getting a much worse version, that is much worse than what you could easily improvise at home. I’m less concerned with ‘EdTech provider [...]
---

First published:

July 8th, 2026


Source:

https://www.lesswrong.com/posts/XA8fMCnwuc45uZYHX/childhood-and-education-20-phones-and-screens

---

Narrated by TYPE III AUDIO.

There is a new very cool Anthropic paper: Verbalizable Representations Form a Global Workspace in Language Models. You can read the blog post verison here.

I encourage reading of the whole original blog post or paper, if you have the time.

Table of Contents


Through A Different Lens.

Establishing J-Space As A Global Workspace.

Are You Pondering What I’m Pondering?

Assistant J.

The Power Of Virtuous Thinking.

High Praise.

Everyone Remains Confused About Consciousness.

Further Research.

Don’t Think.

Through A Different Lens

They call this discovered area of ‘conscious access,’ where things are available for the model to do what in humans we would call conscious reasoning, the ‘J-space,’ after a new interpretability technique called the Jacobian Lens.

The Jacobian Lens computes, for each layer, the average causal effect of changes in the residual stream on the model's eventual outputs, averaged across a wide variety of contexts. Then you can trace what concepts are associated with each layer as the model proceeds through.

At each layer, the J-lens vectors form an overcomplete set.

… We observe that only a relatively small [...]
---
Outline:
(00:26) Through A Different Lens
(02:02) Establishing J-Space As A Global Workspace
(04:55) Are You Pondering What I'm Pondering?
(08:50) Assistant J
(09:42) The Power Of Virtuous Thinking
(17:08) High Praise
(22:37) Everyone Remains Confused About Consciousness
(31:17) Further Research
(34:26) Don't Think
---

First published:

July 7th, 2026


Source:

https://www.lesswrong.com/posts/EnxHPxJT4Xin5cTsX/no-space-like-j-space

---

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.