To scale Perplexity, CEO Aravind Srinivas says large models now label, evaluate, and train smaller ones. The system handles millions of queries daily. Human evaluation is no longer the bottleneck. “An AI has to do this job.”

Perplexity CEO Aravind Srinivas at Harvard Innovation Labs.

Scan any documents, convert images into text, PDF files, etc. 👍

we went from humans labeling cats to this in like 4 years

None of that is new - distilling, generating data and evaluation have all been part of Deep Learning for years now

💯

Still incredibly inefficient to be training new models constantly. It's as simple as giving a model a blockchain and the same model levels up forever with no retraining.

He has a JavaScript omni channel wrapper.

sounds like a fancy way to say they’re just automating the automation. but hey, if it works for certain tasks, cool. just don’t expect it to replace good ol’ human intuition entirely.

This is a profoundly stupid argument. Were humans having to annotate emails to get a spam engine to work? no. Does this auto classification result in actual Classifiers? or just blobs that produced the best result that day?

For whom it may concern: Scientists found a molecule that can cure baldness by waking up dormant hair folicles

this is very cool; every US child gets a $1k S&P account at birth. more stuff like this please!

this story is going wildy viral on reddit. ChatGPT flagged a hidden gene defect that doctors missed for a decade. ChatGPT ingested the patient’s MRI, CT, broad lab panels and years of unexplained symptoms. It noticed that normal serum B12 clashed with nerve pain and fatigue, hinting at a methylation block. Within months tingling eased and brain fog cleared. The primary physician reviewed the genetics report and agreed the variant unified the entire case. IMO, time has already come, taking a 2nd opinion from the best healthcare-AI model should be made part of medical code of practice. ------ reddit. com/r/ChatGPT/comments/1lrmom4/chatgpt_solved_a_10_year_problem_no_doctors_could/

This really hits close to home for me. My whole career in tech was one big unsuccessful struggle to try to explain the distinction between these various modes of operation, and why you will fail miserably if you try to impose a single mindset for all of them.