After Automation

by Katie Parrott in Context Window

Did someone send this newsletter to you?Sign upto have it delivered to your email.

'AI and Me': Increased Automation, Enhanced Human Effort

Currently, we are launching a new episode of our podcast,AI & I. In a reversal of format,Dan Shippersits down with the COOBrandon Gellnot to interview a guest, but to be interviewed himself about why automating tasks results in more human work. The occasion is“After Automation,”Dan's 8,000-word essay about the subject that turned into our most shared article of the year,driving the AI discoursefor a few days on X

It's an unexpected idea from someone leading a company that automates just one task. Yet, it has expanded from four people to 30 during the GPT era, withagents embeddedtoward almost workflow. Dan's argument is not that AI will not transform work—it already has—but that it increases the need for human expertise, judgment, andtaste.

Watch on X or YouTube, or listen on Spotify or Apple Podcasts. You can also read thetranscript.

Here are the highlights:

1. Artificial intelligence enhances the worth of specialists.When someone is able to create a satisfactory initial version—of code, writing, or design—the standard increases, along with the volume of similar content. "You fill the area with lots of material that's near, but not quite accurate," Dan explains. Moving from near to exceptional demands experts who can collaborate with AI to surpass the new standard.
2. The goalposts will continue to shift. Models improve exponentiallyon benchmarks specifically because benchmarks represent fixed frameworks, or established methods of presenting a problem that the model can learn from. Humans are still essential because we can function beyond these predefined frameworks—we step back, reframe the issue, and make unexpected, self-directed decisions that aren't present in the training data.
3. "AI-related job cuts" are often just an excuse.Several tech companies, including Meta and ClickUp, have recently cut staff and attributed the changes to AI. Dan and Brandon see the same pattern: AI serves as a more convenient excuse than acknowledging that the company overhired or is facing financial difficulties. AI will certainly transform the way people work—and large, inflexible organizations will need to adapt accordingly—but this is not the same as the technology removing jobs entirely.
4. Drive the models and you'll be fine.The central contradiction in Dan's essay is that AI generates more tasks for humans while increasing the standards required for that work. Agents are inherently designed to depend on human guidance; without someone determining what is important and how to improve it, they yield average outcomes. According to Dan, to succeed in an AI-driven environment, utilize new models to handle tasks you already excel at, and you will become more valuable than ever.

Miss an episode? Get caught up on Dan's latest discussions with the co-founder of LinkedIn.Reid Hoffman; the group responsible for creating Claude Code,Cat Wu and Boris Cherny; Vercel cofounder Guillermo Rauch; podcaster Dwarkesh Patel; and others, and discover how they utilize AI to think, generate, and interact.—Laura Entis

Signal

The Pope addresses the methods of AI development

When Pope Leo XIV’s encyclical on AI, Magnifica Humanitas, connected to the internet just after 6 a.m. on Monday, the first action I took was to send it to an AI.

I had been eagerly anticipating the Pope's first significant written teaching, like a liberal agnostic secular humanist holding their breath.amateur Bible scholarA knowledge worker in the AI economy. AI, labor, and the Book of Nehemiah, all in one document? I'm not sure there's ever been a more Katie Parrott-style text.

Nevertheless, I let AI take the initial approach. I had Andy, our in-house editorial assistant, use Claude design to transform it into a comic-book infographic containing the essential information for the team...

Become a paid subscriber to to access this section and discover more about:

1. Why the Pope's first AI-related encyclical is not as anti-AI as it initially appears
2. How does the encyclical align with Dan's argument in "After Automation"
3. What questions should a knowledge worker ask before submitting their work to an AI tool?

Subscribe

Click hereto read the entire article

Would you like the complete text of every article in an RSS feed?Become a subscriber, or learn more.