thomas

Just spent the weekend deep-researching and writing about a relatively arcane topic using ChatGPT, Claude, and Perplexity side by side.

Using multiple LLMs and contrasting responses to weed out the occasional hallucination is a massive accelerator of knowledge.

The LLMs even helped me spot a misleading error in one of the authoritative textbooks on the same topic, which was until then my source of truth.

Game changer in how I’ll be engaging going forward with topics I want to discover more seriously.

Aviator, diver, Etherean, nexialist, PhD in aviation ∩ climate science | Host of /diving /ethfinance /eth-staker /geopolitics /science /singapore /thomas

Interested in ZK, FHE, PUFs, hardware, system design, ai safety, and smart contact security.

Topic was the Riemann zeta function and essentially rebuilding Euler’s and Riemann’s lines of thinking from first principles (leading up to the famous hypothesis).

I’d still need a mathematician to proof-read my writing for accuracy, but I’m pretty happy with it, and got to a level of understanding greater than from reading entire books on the topic (such as Derbyshire’s Prime Obsession).

I used the free version of all three LLMs. I’m thinking of upgrading one to the deep research model

What was the topic? Did you use the paid versions of each product?

Pragmatism (Security) at OP Labs.
Prev: ConsenSys & Coinbase.

Attribs: Friendly, curious, introverted, mid-witted.
Chans: /eth-security /maurelian /stoicism




I’ve found ChatGPT’s conversational style to be more natural, as if I was exchanging with a knowledgeable human. Claude was more terse and the LaTex formatting broke down completely there, but not on ChatGPT, making it difficult to parse the replies. I did encounter one misleading hallucination on ChatGPT though.

Are there specific cases where you prefer ChatGPT to Claude or vice versa?

love to see a video of how you're ~efficiently using them in parallel