LogoThread Easy
  • Explore
  • Thread Compose
LogoThread Easy

Your All-in-One Twitter Thread Companion

© 2025 Thread Easy All Rights Reserved.

Explore

Newest first — browse tweet threads

Keep on to blur preview images; turn off to show them clearly

The original tweet actually wasn’t meant to be a joke about Sam Altman not liking it, but upon posting it a minute ago I realize people might naturally take it that way

The original tweet actually wasn’t meant to be a joke about Sam Altman not liking it, but upon posting it a minute ago I realize people might naturally take it that way

Applying the security mindset to everything @PalisadeAI

avatar for Jeffrey Ladish
Jeffrey Ladish
Sat Dec 20 04:46:54
I love METR’s work and quite like Opus 4.5 and nonetheless I do not like my timelines shortening even a little

I love METR’s work and quite like Opus 4.5 and nonetheless I do not like my timelines shortening even a little

The original tweet actually wasn’t meant to be a joke about Sam Altman not liking it, but upon posting it a minute ago I realize people might naturally take it that way

avatar for Jeffrey Ladish
Jeffrey Ladish
Sat Dec 20 04:44:24
I don't want to end up in a situation where the most ruthless AIs - built by the most ruthless AI companies - win

I want extremely helpful, extremely honest AIs that help us make sense of our situation. I do not want competitive AIs which are more powerful than all of us

I don't want to end up in a situation where the most ruthless AIs - built by the most ruthless AI companies - win I want extremely helpful, extremely honest AIs that help us make sense of our situation. I do not want competitive AIs which are more powerful than all of us

Applying the security mindset to everything @PalisadeAI

avatar for Jeffrey Ladish
Jeffrey Ladish
Thu Dec 18 22:03:16
In the short term - before we get to superhuman AI - we need transparency into how models are trained and evaluated. I think Claude is a pretty good AI. I'm glad Anthropic is trying to train Claude to be a helpful assistant rather than a hard-nosed business operator. However...

In the short term - before we get to superhuman AI - we need transparency into how models are trained and evaluated. I think Claude is a pretty good AI. I'm glad Anthropic is trying to train Claude to be a helpful assistant rather than a hard-nosed business operator. However...

I don't want to end up in a situation where the most ruthless AIs - built by the most ruthless AI companies - win I want extremely helpful, extremely honest AIs that help us make sense of our situation. I do not want competitive AIs which are more powerful than all of us

avatar for Jeffrey Ladish
Jeffrey Ladish
Thu Dec 18 22:03:16
"Claude is trained to be helpful, meaning it’s often inclined to act more like a friend than a hard-nosed business operator"

This is cute now but I will refer back to this tweet once it becomes extremely ominous

"Claude is trained to be helpful, meaning it’s often inclined to act more like a friend than a hard-nosed business operator" This is cute now but I will refer back to this tweet once it becomes extremely ominous

My main concern is that humans won't know how to align superhuman AIs at all. But I think this challenge will be made even more difficult due to incentives to make your AI a "hard-nosed business operator" I expect competitive pressures to reward companies who do this

avatar for Jeffrey Ladish
Jeffrey Ladish
Thu Dec 18 22:03:15
Claude is a way better name from a marketing point of view, and I’m pretty unsure if this is net good or net bad for the world

Like I think it’s also kinda good Gemini is a mouthful and Grok sounds so weird 

Claude is maybe too normal

Claude is a way better name from a marketing point of view, and I’m pretty unsure if this is net good or net bad for the world Like I think it’s also kinda good Gemini is a mouthful and Grok sounds so weird Claude is maybe too normal

Ani… I don’t like Ani

avatar for Jeffrey Ladish
Jeffrey Ladish
Thu Dec 18 20:38:48
  • Previous
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • Next