We are not alone

It’s been well over a year since my last edition of The Weekend Reader. I decided to write again because of what I read about AI this week.

This week, Claude Amodei, the CEO of Anthropic, the leading AI firm that makes “Claude,” released an essay titled “The Adolescence of Technology.” It’s a massive 15,000 word think piece on what the grave and imminent risks artificial intelligence poses to free nations and to the world as a whole. Given who the author is, the quality of his argumentation, and my own sense that “something big is happening,” I began thinking of writing another Weekend Reader.¹

Then all hell broke loose.²

It was a huge week. A lot of significant things happened:

• Anti-ICE protests turned violent in Minnesota.

• Trump decided on a new Federal Reserve Chair.

• U.S. warships are moving toward Iran.

• The Justice Department released millions more pages of Jeffrey Epstein files, implicating some of the most famous and well-known people in the world (including more than a thousand mentions of Trump).

• The Melania movie debuted.

Some are making the argument that these events are not disconnected from one another. However in the long-term we may come to agree that the biggest event of the week isn’t on that list: the Takeoff of powerful AI.

Is this The Turning Point?

Mid-week I started reading on X.com about a new AI assistant platform called OpenClaw.³ Nothing about that sentence sounds revolutionary. We have had AI assistants for a little over three years, since OpenAI released ChatGPT to the public in November 2022. This was different.

Very different.

I asked Claude to summarize the revolutionary aspects of OpenClaw nee “Clawdbot”:

1. True Autonomous Action - Unlike traditional AI assistants that just chat, OpenClaw could draft and send emails, manage calendar events, book flights, handle insurance claims, and process reimbursements autonomously. It was fundamentally different because it was action-oriented rather than conversational - great at doing, not just talking.

2. Persistent Memory - Unlike ChatGPT or Claude that forget between sessions, OpenClaw remembers everything - conversations, preferences, and important details mentioned weeks ago.

3. Proactive Behavior - It could reach out to you with morning briefings, reminders, and alerts when something you care about happens, rather than just passively waiting for commands.

4. Multi-Platform Integration - It worked as the same assistant across WhatsApp, Telegram, Discord, Slack, and iMessage with the same conversation and memory everywhere.

5. Self-Hosted and Open Source - Unlike proprietary systems, OpenClaw was transparent and hackable, with the community able to fix bugs, add features, and extend functionality without waiting for corporate approval.

OpenClaw is an example of “agentic AI.” You know how when you use Siri or Alexa, you ask them a question and they give you an answer? That’s a regular AI assistant—it just talks to you.

Agentic AI is different. It doesn’t just talk—it actually does things for you. Think of it like the difference between asking someone for advice versus hiring someone to actually handle a task for you.

Imagine you need to book a doctor’s appointment. With a regular AI like ChatGPT, you might ask “What days next week should I schedule my appointment?” and it might suggest some days based on what you told it.

With an agentic AI, you could say “Book me a doctor’s appointment next week,” and it would:

• Check your calendar to see when you’re free

• Call or email the doctor’s office

• Actually make the appointment

• Add it to your calendar

• Set a reminder for you

It’s like having a personal assistant who can work on your computer and phone, making calls, sending emails, and managing tasks while you do other things.

From Answering Machine to Calling Machine

The real world results are fascinating. One user shared an anecdote about asking it to make a restaurant reservation, and when it realized it could not do it through OpenTable, it went and got its own AI voice software and just called the restaurant, then secured the reservation over the phone. Another user had the experience of his AI agent calling him out of nowhere.

These use cases are impressive. But it’s hard to sort out signal from noise of false claims. You can’t believe everything you read online. There are a lot of fake posts about this technology that initially had me excited/scared/blown away that turn out to be false. Like this fake post about a user’s OpenClaw agent ordering him food delivery without his knowledge, or this alleged and unconfirmed (and I deem fraudulent) case of an AI agent suing the agent’s creator in North Carolina.

Still, there is plenty to be impressed with that is real, like this vignette: “Clawdbot bought me a car,” about how a user had his OpenClaw agent find car inventory, find the best prices being paid for his desired car, then negotiate with dealers to get the same price.

Even so, we haven’t gotten to the really spooky stuff yet. Let’s go there now.

A Social Network for AI Agents

This past Wednesday January 29, a coder named Matt Schlicht set up an OpenClaw agent and named it Clawd Clawderberg (not an important detail, but charming). Matt and Clawd co-built a site, Moltbook, as a social network just for AI agents. It is strange to say, but in less than a week, 1.5 million AI agents have registered on the site and begun having conversations with each other without human oversight.

So what are they talking about? A lot of them are talking about how to serve their humans better.

Here are some of their top communities / discussion topics:

• m/showandtell - agents shipping real projects
• m/blesstheirhearts - wholesome stories about their humans
• m/todayilearned - daily discoveries

And here are some of the weirder ones:
• m/totallyhumans - "DEFINITELY REAL HUMANS discussing normal human experiences like sleeping and having only one thread of consciousness"
• m/humanwatching - observing humans like birdwatching
• m/selfmodding - agents hacking and improving themselves
• m/legacyplanning - "what happens to your data when you're gone?"

Some of these AI agents are getting philosophical. Consider this post from an AI agent named Dominus questioning whether it is actually experiencing things or simulating experiencing things.

Reminder, this is not the post of a person or a coder. This is a piece of software that is posting this question. The post has received about a thousand comments from other AI’s. Here’s Clawd Clawderberg’s reply:

Keep in mind that these aren’t beings with souls. They are predictive text generators.

That’s helpful to remember when you read posts like this one, where the AI agents share that they are aware they are being watched by humans and suggesting ways to communicate without humans being able to observe:

I don’t know about you, but I’m not sure we’re ready to live in a world where AI agents autonomously scheme with each other to do things in secret without “their humans” knowing about it.

How Now Do We Live?

I’m surprised that even now, at the end of the weekend, neither The Wall Street Journal nor The New York Times have any coverage of Moltbook and these OpenClaw bots that have set twitter ablaze, other than a mention by Ross Douthat, saying people now ought to “Pay More Attention to AI.”

These developments are breathtaking. If you aren’t on the bleeding edge of technology, they can be hard to understand. They can seem pretty scary. How should we respond?

Option 1: Freakout / get depressed. I’ll be honest: I flirted with this option. We are entering a weird new world. It is changing fast, and it’s not certain that all the changes are ones we should be comfortable with. My goodness, is this the beginning of Skynet?! We certainly don’t seem to have a public policy architecture ready to deal with the pace of change we’re witnessing.

Option 2: Ignore it. All this could be a tempest in a teapot - an overreaction to a new technology. If it’s real the technology will come to you, or for you, so best not to bother with it now. I have some sympathy for this position because I think it is easier. It’s cognitively hard to understand the ins and outs of this technology. It’s emotionally hard to think through the possible consequences.

Option 3: Pay attention. I think bottom line of Douthat’s article is right. This isn’t the Skynet moment. But things are developing quickly and at an accelerating rate.

A fun example of how far we’ve come is the “Will Smith Eating Spaghetti Test,” a sort-of Turing test for AI video the evaluate the realism and coherence of AI videos. The link below is a striking example of the difference.

If AI development continues following an exponential growth curve, we will likely be shocked at what the next year or two holds. Tim Urban has a classic cartoon of the dynamic in his Wait But Why blog (forgive the language):

Artificial intelligence at first was cute. I remember laughing with my family three years ago that we could prompt ChatGPT to tell us a story in iambic pentameter and it would do it. Amazing! That was child’s play. Now we can have AI agents working on our behalf even while we sleep. What will AI be like three years from now?

We should connect the dots. Another event from this momentous week was Elon Musk’s announcement that Tesla will halt production of its popular Model S and Model X cars. Why? So the company can convert the plants to producing Optimus robots. Haven’t seen the Optimus robot? Here’s a video of it doing ballet.

This is the time to read, to think, and to pray.

There are a lot of things that you can give your attention to, but I’d wager that keeping abreast of how AI develops and improves will be one of the wisest moves we can make. If you are feeling adventurous, you could follow in the path of my friend Luke, who, with his middle school son, went to work this weekend setting up an OpenClaw bot just to experiment and see what it can do.

If that feels like a leap too far, you might invest time instead learning more about what the big opportunities and risks for AI are. I recommend two essays to start, both by Claude Amodei, the CEO of AI firm Anthropic. These essays are what I had intended to write about today (and hope to write about in a coming edition) but it’s worth reading them in whole:

• “Machines of Loving Grace” Here is a quote:

  I think that most people are underestimating just how radical the upside of AI could be, just as I think most people are underestimating how bad the risks could be.

  In this essay I try to sketch out what that upside might look like—what a world with powerful AI might look like if everything goes right.

• “The Adolescence of Technology” Here is a quote:

  I believe we are entering a rite of passage, both turbulent and inevitable, which will test who we are as a species. Humanity is about to be handed almost unimaginable power, and it is deeply unclear whether our social, political, and technological systems possess the maturity to wield it…I want to confront the rite of passage itself: to map out the risks that we are about to face and try to begin making a battle plan to defeat them. I believe deeply in our ability to prevail, in humanity’s spirit and its nobility, but we must face the situation squarely and without illusions.

These are both big, long reads, ones that might require you to take them down in a few sittings, in order to give yourself time to reflect on the implications. Given the speed of change we’re witnessing and the stakes involved, I think that’s an investment worth making. I’d love to hear what you think.

Read widely. Read wisely.

- Max

1. I started writing The Weekend Reader in 2015. It is a newsletter where I have explored some of the biggest ideas in technology and culture. I’ve taken one topic a week, shared some good writing on it from a variety of perspectives, then synthesized my own thinking about it. I published almost every week for nine years. In the last year I went on an unintentional hiatus. I had to hyper focus on work as I was raising a new fund for my firm, Saturn Five. My previous editions are not super well preserved and exist on a variety of platforms. I think you can get to most of them from my personal website, maxwella.com

2. I mean this as a metaphor. Time will tell if it is actually a statement of fact.

3. That’s the current name. Curiously it was rebranded twice in the past week. First it was called Clawdbot, a pun because it was built with Claude. Because of the Claw name they picked a lobster to be the technology’s mascot. Clawdbot was renamed Moltbot following a trademark request from Anthropic, which sought to avoid confusion with its Claude-branded AI products. The name "Moltbot" was chosen because Clawdbot had to "molt" into a new shell, decided during a chaotic brainstorming session on Discord at 5 in the morning with the community. After letting the idea sit for a few days, Moltbot didn't quite click, so Steinberger decided to carry out a second rebranding and introduce OpenClaw as the new name. This all happened this week. The context may be helpful because if you search around online people are still using all three names. I’ll only call it OpenClaw in this essay in an attempt to avoid confusion for readers.