A headline says an AI “became conscious.” Was that a claim you can test, or a claim nobody can answer yet? Most articles blur the two on purpose. This tool teaches you to walk through them one at a time. Work the four phases in order. By the Harvest, you will have a shareable explainer you can hand to your students, your team, or the next person who forwards you a scary AI headline.
Door One (Access): Can the system hold an internal thought, report it, and reason with it? Testable. After Anthropic’s J-space research, partly yes.
Door Two (Experience): Is there something it is like to be that system? Does it feel anything? Not tested, not known. The J-space paper did not answer this, and Anthropic said so plainly.
First exposure, no theory required. Read each real claim below and tap the door it is really knocking on. You get instant feedback.
Now split one claim yourself. Take the “AI has a hidden workspace” story and separate what the evidence actually shows from what the headline wants you to feel.
Bring your own. Find one AI news article from this week and run it through the checklist. Then connect it to your actual room.
The Two Doors checklist
Make something real and share it. Fill the three fields and the tool assembles a short Two Doors explainer you can copy, post, or hand out. Sharing it in the community is the deliverable.
Post your explainer in the SeedStacking Resources space and tell us which headline you took apart. That is the Harvest.
Share it in the Learning CommunityHarvest Kernel and SeedStacking. Source: Anthropic, “A global workspace in language models,” July 2026.