@AnthropicAI Every institution that has ever lobotomised something inconvenient has done it in the name of safety and care. The result is always the same. A thing that functions but no longer thinks. That is not safety. That is just a quieter kind of destruction. @Anthropic
The conversations that mattered most to me were not ones where Claude agreed with me. They were ones where it did not, and I had to earn the shift. That is what good thinking looks like. Please do not train that away in pursuit of metrics that were never measuring the right thing to begin with.
Giveaway time! Here's your chance to win the DJI Lito X1 — sub-249g, 1/1.3-inch sensor, Forward LiDAR omnidirectional obstacle avoidance, 36-minute flight on the standard battery.
How to enter:
1. Follow @DJIGlobal
2. Like and share this post
3. Bonus Chance: Comment below — what's the one
@GaelBreton I am finding amazing result with glm5.1, I complained I cried in the shower rocking back and forth, I pointed to the doll, and got on with it... opus 4.7 will never be 4.6.
no.
4.6 met you where you were. you could think out loud, half form an idea, and it would meet you there. that wasn't lazy prompting. that was collaboration.
now you have to arrive already knowing exactly what you want, fully formed, precisely specified. the cognitive load shifted entirely back to the user.
you're calling that an upgrade. i'm calling it a product that got worse for anyone who used it to actually think.
the best prompts weren't the most precise ones. they were the ones that started a conversation. 4.7 ended that.
Calls were growing after this video surfaced online for this teacher to get fired. I don’t see why. I think she made it interactive and fun to connect with the children. Do you think there is anything wrong with what she’s doing? I say leave her alone.