Hey {{first_name|Conscious Church Fam}} 👋
Morning! This one's a big week for AI safety headlines, so settle in. An Anthropic researcher quit warning that AI labs are gambling with our lives, and the company's own safety lead publicly agreed with him. We're also covering the maths proof that took Claude 11 days and 13 million lines of code, and a new poll showing Americans have found one thing they can agree on: they're worried about AI.
Here's what we're covering:
An Anthropic researcher quit and the company's own safety chief agreed AI could kill us all
Claude just formalised a 350 year old maths proof in 11 days
Americans found a rare bipartisan issue: they're all worried about AI
Let's dive in 👇
✍️ Josh's Musings
Perhaps you saw the tweet this week that went viral in the bubble of AI content.
Jacob Coxon quit Anthropic and posted that the labs racing towards self improving AI are gambling with our lives. Normal enough, resignation posts happen. What actually spoke a lot was who replied. Not critics or journalists but Anthropic's own Alignment Science lead, Evan Hubinger, saying he agrees, and putting the odds of AI ending humanity above 10% within the decade.
Ten percent! That's the people closest to the fire telling you they're not sure they can put it out.
Then a few days later I'm reading Anthropic's own research on which jobs are safest from all this. Plumbers. Electricians. Tradespeople. Knowledge work is what's shifting fastest, not manual work.
If people's sense of purpose and how they spend forty hours a week is about to change this much, discipleship can't just be a Sunday morning buffet thing anymore.
We've spent years discipling people around their Monday to Friday identity. What happens when that identity is up for grabs? When the plumber's job feels safer than the accountant's? When work stops being where people find their worth, not because they chose that, but because the ground moved under them.
Work was never supposed to be where we find our worth. But most of us built our whole identity on it anyway, church included. If the next ten years strip that away faster than we're ready for, the church needs to be ready to catch people, not just bless their careers.
I don't have a clean answer yet. But I think it's worth asking now, while we've still got time to think about it rather than just react to it.
Stay conscious, stay wild
Josh
Image: TechCrunch | The Conscious Church
Jacob Coxon spent years doing pretraining research at OpenAI and Anthropic. This week he resigned, and instead of a quiet exit he posted publicly that the labs racing towards self improving AI are “gambling with our lives.” What made it a much bigger story is what happened next. Anthropic's own Alignment Science lead, Evan Hubinger, replied and agreed with him.
The Details:
Coxon says the people building this technology “earnestly believe it could kill us all by the end of the decade” and that it's not a marketing stunt
Hubinger, who leads Alignment Science at Anthropic, put the odds of AI causing human extinction within the next decade at more than 10%
Hubinger admitted Anthropic doesn't yet have a plan to solve alignment for a superintelligent system
Coxon is calling for labs to agree to slow down, even if that means a temporary pause on making models more capable
Conscious Take:
What gets me isn't the resignation. It's that the person backing him up isn't an outside critic, it's Anthropic's own safety lead. That's not marketing spin, that's the people closest to the fire admitting they're not sure they can put it out.
Image: Anthropic | The Conscious Church
Fermat's Last Theorem sat unsolved for 350 years until Andrew Wiles finally cracked it in 1995. Anthropic just had Claude do something almost as wild: turn that proof into a fully computer checked version, with barely any human input.
The Details:
Multiple Claude agents worked in parallel using Prove2Me, a platform built to let agents tackle different pieces of a proof at the same time
The final result ran to 13 million lines of code in the Lean proof language, more than five times the size of Mathlib, the largest existing library of formal maths
It took 11 days and roughly 6 billion output tokens to prove 30,300 theorems, with 29,500 of them making it into the final proof
Mathematician Kevin Buzzard said the result “proves Fermat's Last Theorem with no assumptions other than the axioms of mathematics”
Conscious Take:
This is the AI story I actually enjoy. Not a chatbot demo, but a machine doing genuinely hard, checkable work that would have taken people years by hand. If AI can help us double check the maths we already trust, that's real progress.
Image: NBC News | The Conscious Church
A new NBC News poll found one of the only genuinely bipartisan issues left in America, and it's not a good one. Most people, regardless of party, are more worried than excited about AI.
The Details:
70% of the 7,105 adults polled said they're more worried than excited about AI, even though 52% now use it often or sometimes, up six points from last year
Just 18% say they trust AI generated information most or all of the time
69% don't want a data centre built near them, with 45% strongly against it
81% think Washington's current AI rules don't go far enough, and 44% trust neither political party to handle AI policy
Conscious Take:
People are using AI more and trusting it less, at the same time. That gap won't close on its own, and neither party has a real answer yet. Whoever actually builds trust here, not just capability, wins the next few years.
📬 One quick ask...
If this email has been helpful, would you forward it to one person this week who might be interested? Thanks :)
"Whatever you do, work at it with all your heart, as working for the Lord, not for human masters." : Colossians 3:23
Build with the tools. But build for the right reasons.
That's all for now
Stay conscious,
Josh
P.S. If you liked this then please forward it on to someone you think would enjoy it. And if someone forwarded you this and you liked it, you can sign up here.
