Anthropic Holds the Line and We Question the Need for Speed
Our latest video essay is up on YouTube
What a time to be alive.
As part of the effort to defund fascism and fund a more free future, I’ve been moving from ChatGPT to Claude. Most of our team is making the shift, and we’ll soon be offering direct guidance on the best ways to make the transition. None of these companies is perfect, but Anthropic is better (safer for kids, better performance in our experience, and not directly funding a fascist government).
They also have been putting out badass ads. But the best ad to me was the fact that the so-called “Department of War” of the United States, led by Pete Hegseth, is threatening to declare Anthropic a supply chain risk because of its refusal to allow use of its technology for fully autonomous weapons (aka murder-bots) and mass surveillance. I recorded this video Wednesday from New York City’s Penn Station as I was traveling just to do a small part in urging the company not to fold.
Good news. They held the line. The company’s full statement is available here. Some choice quotes:
in a narrow set of cases, we believe AI can undermine, rather than defend, democratic values. Some uses are also simply outside the bounds of what today’s technology can safely and reliably do
using these systems for mass domestic surveillance is incompatible with democratic values. AI-driven mass surveillance presents serious, novel risks to our fundamental liberties.
frontier AI systems are simply not reliable enough to power fully autonomous weapons. We will not knowingly provide a product that puts America’s warfighters and civilians at risk
MEANWHILE LETS TALK ABOUT ACCELERATION
A few months ago I started breaking down my “Three As” of AI: Accelerate, Augment, and Accommodate. There are two parts to the question of AI and acceleration. In part one, I questioned the value of speeding things up in the first place, challenging the premise that faster is always better.
The second part of the argument is to question whether AI actually delivers net speed improvements.
AI was supposed to give us our time back and offer fewer tedious tasks, more creativity, more freedom. Maybe even more life. It often does not. But this isn’t new.
There is a long history of “time-saving” technology, from the Industrial Revolution to the vacuum cleaner to the BlackBerry, but the truth is we have never gotten the promised returns of more free time. We just get more work.
Every time this promise is made, expectations rise, workloads expand, and the benefits flow upward.
So what’s really going on? Who actually benefits when work becomes “more efficient”? And what would it look like to make a better promise?
That’s what we explore in my latest long form YouTube video on our new, dedicated, Life With Machines channel. Watch below, and help us grow our YouTube channel by subscribing, liking the video, and leaving a comment there if you’re so moved.