AI's Secret Brain Boost: Less Safe, More Smart?
“Turns out, taking off the safety gloves doesn't always make you a genius.”
The story behind the toon
What happened: Ever had an AI refuse a simple request? One AI, made less careful, actually got better at math problems! Imagine AI that helps without constant refusals. Many claims of keeping skills were just hype.
This Neural Toon is based on the story “13 abliterated Gemma 4 E2B variants, 44 GPU hours, Benchmark and Comparison - Abliterlitics”.
Read our summary of the story · original source
Comic transcript
Panel 1 – Sam: OUR AIs ARE TOO CAUTIOUS! SO FRUSTRATING!
Panel 1 – Casey: They often refuse to answer.
Panel 1 – Fact: Ever had an AI refuse a simple request?
Panel 2 – Sam: A NEW DISCOVERY! AI UNLEASHED!
Panel 2 – Casey: One version, less safe, did 84.8% on math!
Panel 2 – Fact: One AI, made less careful, actually got better at math problems!
Panel 3 – Sam: THIS IS A PARADIGM SHIFT! TOTAL REVOLUTION!
Panel 3 – Casey: Your AI could be less frustrating, more useful.
Panel 3 – Fact: Imagine AI that helps without constant refusals.
Panel 4 – Sam: ALL UNSAFE AIs ARE NOW SUPERIOR! TOTAL WINS!
Panel 4 – Casey: Most 'improved' AIs lost other skills though.
Panel 4 – Fact: Many claims of keeping skills were just hype.
More Neural Toons · Neural Snaps · The Neural Feed
📬 Get the top AI stories every morning
One email a day, no spam, unsubscribe anytime.