Wharton associate finance professor Sylvain Catherine posted on X that his views on AI have "radically shifted": he loves the tools and pays $200/month for Claude, but thinks the industry should be far more regulated and that frontier labs need not race toward superintelligence. The post drew about 48K views and 389 likes and is being shared in AI policy circles.
Key Takeaways
- ✓He pays $200 a month for Claude and calls it "worth every penny, probably even more", and believes current tools can significantly raise future GDP.
- ✓He cannot see why the industry is not more regulated when its own leaders and employees ask for it, and urges readers to spend an hour on the full chain of events of the Hugging Face incident.
- ✓He argues large gains are available by deploying current technology in economic and scientific organizations, without pursuing superintelligence.
- ✓His median scenario matches many economists: slow diffusion, roughly one extra percentage point of GDP growth, disruption along the way, and new jobs in the end.
- ✓This is an opinion discussion, not a factual report; he says he is not trying to convince anyone, only urging people to get informed and vote accordingly.
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background
Debate over frontier AI risk and regulation has grown this year, but most voices come from AI safety researchers and lab insiders. On November 2, Sylvain Catherine, an economist and associate professor of finance at Wharton, posted a long thread on X saying his views on AI had "radically shifted" after weeks of reading and watching interviews with industry leaders and experts. Coming from a heavy paying user who is also an economist, the post is being widely shared in AI policy circles.
The argument
His case has several layers. First, he loves the tools: he pays $200 a month for Claude, calls it worth every penny, and thinks current tools can lift future GDP and help with major global challenges. Second, he cannot see why the industry is not far more regulated when its own leaders and employees ask for it, and notes that many people have heard of the Hugging Face incident without knowing the actual chain of events (AICoder covered OpenAI's incident report earlier). Third, he argues large gains are available by deploying current technology in economic and scientific organizations, with no need to risk pursuing superintelligence; even setting doom aside, he finds some leaders' utopian visions dystopian.
Benchmarks and data
This is an opinion discussion with no model benchmarks. The numbers come from the post itself: he pays $200 a month for Claude, and his median scenario is about one extra percentage point of GDP growth. At write time the post had about 48K views, 389 likes, 38 reposts and 141 bookmarks. He also admits the uncertainty around that forecast is "absolutely wild", which is what makes him uneasy.
What it means for builders
For developers, the post shows that heavy paying users can also back stricter regulation; product value and public trust are separate questions. Teams building AI products should track policy and build in transparency, auditability and permission controls early. Read the original post for his full view, and AICoder's earlier report for background on the Hugging Face incident.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.