Spent months building a living scorecard for SaaS & Stablecoin/Web3 core banking vendors. 46 platforms each, sourced, weighted, updated quarterly.
Opening early access today. Want the link? Comment or DM me.
Say whatever the fuck you want
Something broke in crypto on 10/10
People want to minimize this but the market simply does NOT function like it did for literally YEARS prior
We are seeing massive bull markets in metals, AI, space stocks etc
NOTHING makes its way to crypto ...
We are getting ransacked and we deserve answers
🚨 RIP prompt engineering.
This new Stanford paper just made it irrelevant with a single technique.
It's called Verbalized Sampling and it proves aligned AI models aren't broken we've just been prompting them wrong this whole time.
Here's the problem: Post-training alignment causes mode collapse. Ask ChatGPT "tell me a joke about coffee" 5 times and you'll get the SAME joke. Every. Single. Time.
Everyone blamed the algorithms. Turns out, it's deeper than that.
The real culprit? 'Typicality bias' in human preference data. Annotators systematically favor familiar, conventional responses. This bias gets baked into reward models, and aligned models collapse to the most "typical" output.
The math is brutal: when you have multiple valid answers (like creative writing), typicality becomes the tie-breaker. The model picks the safest, most stereotypical response every time.
But here's the kicker: the diversity is still there. It's just trapped.
Introducing "Verbalized Sampling."
Instead of asking "Tell me a joke," you ask: "Generate 5 jokes with their probabilities."
That's it. No retraining. No fine-tuning. Just a different prompt.
The results are insane:
- 1.6-2.1× diversity increase on creative writing
- 66.8% recovery of base model diversity
- Zero loss in factual accuracy or safety
Why does this work? Different prompts collapse to different modes.
When you ask for ONE response, you get the mode joke. When you ask for a DISTRIBUTION, you get the actual diverse distribution the model learned during pretraining.
They tested it everywhere:
✓ Creative writing (poems, stories, jokes)
✓ Dialogue simulation
✓ Open-ended QA
✓ Synthetic data generation
And here's the emergent trend: "larger models benefit MORE from this."
GPT-4 gains 2× the diversity improvement compared to GPT-4-mini.
The bigger the model, the more trapped diversity it has.
This flips everything we thought about alignment. Mode collapse isn't permanent damage it's a prompting problem.
The diversity was never lost. We just forgot how to access it.
100% training-free. Works on ANY aligned model. Available now.
Read the paper: arxiv. org/abs/2510.01171
The AI diversity bottleneck just got solved with 8 words.
Some say digital finance is the future, I say it's already here.
BLACKROCK
- 19,800 employees
- $10.4T AUM (10% of world GDP!)
- $5.5b profit
TETHER
- approx 50 employees
- $119B AUM (0.1% of world GDP)
- $6.2b profit
I spoke with one of Tether's founders, here's 2 insights.
By 2034, your 9-5 job will be extinct.
Everything will be so cheap you won't have to work anymore.
That's Marc Andreessen's latest prediction — the billionaire investor who predicted the rise of the internet in 1993.
Here's what he said next:
BREAKING:
Ethereum and Solidity are now formally integrated into high school education programs in Buenos Aires
High schools students will learn to write #Ethereum smart contracts
Vamos Argentina! 🇦🇷