Abacus.AI's Bindu Reddy claims Fable AI fails outside of coding, while researcher j⥉nus blames poor user operation

@bindureddy so is it really as insane that the government had to ban it?!

@bindureddy I feel like you gotta be seriously failing to know how to use it and yield it, to a degree that’s rare, if you think it sucks for everything else

@bindureddy And out of credits before you figure it all out.

@BenHolfeld No - the ban was totally random

@bindureddy that’s a fair trade-off for high-level complex tasks tbh

@bindureddy and it should not be used to power Abacus, two steps backwards yesterday.

@syedmuzayan well.. so use smartly

@bindureddy Narrowly strong tools like Fable need a precise input gate. Without one, they treat every query as complex, leading to overengineering and hidden errors. How do you isolate its hard-coding domain?

@bindureddy Fable is so underated for what it can do with UI and UX design

@bindureddy lol false, you could not be farther from the truth

@bindureddy Do you have some examples you can post? Really curious to see where it's falling short.

@bindureddy like running a scalpel when everyone expects a swiss army knife

the skill floor is the feature, not the bug

@bindureddy Fable's narrow coding focus suggests AI will keep excelling where its creators put effort.

i think you mean wield**

its even better if you learn how to collaborate with them, instead of handling "it" as if it were a hammer in your hand.

treat fable like a dancing partner, or a teammate, or a coworker. your experience will 10x+

@bindureddy guess the government needs to figure out how to handle this.... let's see if anyone is gonna jailbreak it... if its really good at coding and hacking maybe it can break into some systems that regular human hackers can not

@bindureddy what had changed? do they still train using Transformer Architecture? may not be due to high performance GPU; maybe good quality datasets!

@bindureddy wat? it rejects literally everything I give it

@bindureddy what it sucks for?

@bindureddy Capability is becoming increasingly task-specific. Matching the model to the problem often matters more than chasing the highest benchmark.

@ChainZenit @bindureddy High-level, as opposed to making videos showing cats sprouting wings and flying? 😹