Prompting a frozen model is nothing like reasoning with a human
Comparing prompting a frozen model to reasoning with a human, there is absolutely no comparison — the two are qualitatively different activities.
Explore
31 results · Most relevant
Results
WHY · 30 of 33209 results · federal reserve
Open question: how much edge can a machine actually capture?
The unanswered question: if real edge lives in a human brain, how much can a machine actually capture? Whoever answers that is not building a product — they're building a new way of thinking in finance.
explicitly framed as unanswered
Edge is built by sitting next to great traders — and it walks out the door with them
Investors get their edge through decades of working and learning from some of the best traders in the industry by sitting next to them. But when an investor whose knowledge lives entirely in their head moves firms or retires, that intellectual property leaves with them.
described with high confidence
What financial investors actually do all day
Financial investors keep up with markets that now move 24-7, try to understand billions of different data connections and how events relate to each other, and compare and understand the historical significance of events happening right now — what happened in the past, why it happened, and what the market did when it happened. A lot of work takes place before a financial decision is actually made.
described with high confidence
If everyone builds the same tools, differentiation is the question to answer
If everyone is building the same tools, what actually makes a tool different is the question builders must answer.
Deep dive into one thing with a 360-degree view of information
Investors benefit from the ability to deep dive into one thing with a 360-degree view of information they wouldn't have had before.
An agent that only apologizes undermines accountability a human provides
Relying on an agent that just apologizes to you instead of a human who can stand behind their decision undermines accountability.
Use an LLM as a debate partner for trading decisions
Use an LLM as a debate partner to talk through and reason about trading decisions, e.g. asking whether FOMC is today or tomorrow.
Weekend tools vs 24-7 infrastructure, and understanding the investor first
A tool built on the weekend differs from one built 24/7 on every market with big data that is reliable and trustworthy — the latter becomes infrastructure. Before building AI into investing, understand what a financial investor is actually doing first. Pulling data from an LLM means giving it context on an old model.
one atom has unresolved attribution and is treated cautiously
Anekha Sokhal's path from derivative trading to ML research
Anekha Sokhal spent nearly a decade as a derivative trader before moving into data science and machine learning, earned a Fulbright scholarship to study ML, and published a paper at ICLR while working on machine learning systems with NASA.
anecdotal, high confidence
Protocols you can run
WHY · 0 of 59 published protocols
No published protocol matches this selection.
Source analyses
WHY · 1 of 627 current source analyses
People worth following for this
The creators behind most of what cleared.
WHY · 1 creator you don't follow account for 30 cleared insights
- AI Engineer
30 cleared insights match
Choose sources → to get a library of your own and a daily brief.