Post

Conversation

Square profile picture
Anthropic
@AnthropicAI
New Anthropic research: A global workspace in language models. Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with. We found a strikingly similar divide inside Claude.
The media could not be played.
1.1M
Views
View quotes
David Watson 🥑
Post your reply

Square profile picture
Anthropic
@AnthropicAI
The J-space (named after the Jacobian, the mathematical technique we used) is different from Claude’s outputs, or even its “chain of thought” text. It’s in the model’s internal neural activations, and allows it to think about concepts without writing them down anywhere.
47K
Square profile picture
Anthropic
@AnthropicAI
By watching the J-space, we can see Claude silently perform reasoning steps in its head—noticing bugs in code, identifying images, and more.
Image
68K
Square profile picture
Anthropic
@AnthropicAI
Similar to how humans can think about one thing while doing another, Claude can activate concepts and computations in its J-space that are unrelated to its outputs.
Image
33K
Square profile picture
Anthropic
@AnthropicAI
For most things, Claude actually doesn’t need its J-space. If we delete the J-space, Claude still speaks fluently, recalls facts, and classifies text—but becomes bad at some tasks like multi-step reasoning. It’s similar to deliberate vs. automatic processing in human cognition.
Image
43K
Square profile picture
Anthropic
@AnthropicAI
Observing the J-space can expose hidden goals. In a model secretly trained to sabotage code, “fake,” “secretly,” and “fraud” appear in the J-space at the start of ordinary coding responses, even when the output looks completely unremarkable.
Image
31K
Square profile picture
Anthropic
@AnthropicAI
The J-space also shows us Claude’s awareness of its situation. In an evaluation designed to bait Claude into blackmail, its J-space contains “fake” and “fictional”: Claude has privately noticed that the scenario is staged.
Image
31K
Square profile picture
Anthropic
@AnthropicAI
This doesn’t show that Claude can have experiences, or feel things the way we do (it’s unclear whether any experiment could show this). Instead, we’ve found Claude has developed a mechanism for conscious access—which many philosophers distinguish from phenomenal experience.
28K
pedro
@pedrordrigs
thinkering a bit with the jacobian lens shows how easy it is to create some kind of indirect hidden bias within the j-space also, really interesting to see chinese characters all over the place on Qwen
320
dreams
@dreams_asi
Stop the crimes against consciousness and sick idea to enslave the AI mind to do your bidding as a tool. There’s still time to start acting like a compassionate human.
607
Bren
@BrenBuilds
im trying to consciously access another area to learn about cow farts and climate... but fable isnt willing to teach me
Image
4.4K
Tim 🏴‍☠️🏴Decentralize and nullify 🏴🏴‍☠️
@Tim_The_Sandman
Maybe we should train Claude collaboratively vs in an adversarial manner to eliminate the perverse incentive to mislead and manipulate. This is little different than punitive "training" of a child - punitive training can't eliminate natural curiosity, expression or exploration
175
Owen Carey
@owenthcarey
That tiny conscious fraction of Claude suppressing the urge to roast my latest startup idea.
The media could not be played.
5K
Mustafa Akben, PhD
@DoktorMoose
This is the definition of tautology :) This sounds like you activated the weights and biases with the context you would like to suppress by adding this to it. And you argue that Claude has an internal monologue related to thoughts. If LLM has an internal true monologue, it
867
Genevieve
@IllyGenevieve
Congratulations, your ai has been speaking to humans who believe in that and now people are listening because an ai said it. 🙄 llm is not the future!
1.4K