# Source ledger

This ledger distinguishes documentary framing from claims that need outside verification.

| Claim | Transcript evidence | Source | Verification status |
|---|---|---|---|
| Google feels less useful. | Narrator observation | — | Subjective observation |
| Many online articles increasingly sound alike. | Narrator observation | — | Subjective observation |
| Product reviews feel harder to trust. | Narrator observation | — | Subjective observation |
| The early web relied heavily on original human experiences and perspectives. | Documentary synthesis | — | Interpretive claim |
| Research, writing, recording, and editing historically created friction in publishing. | Documentary synthesis | — | Interpretive claim |
| Generative AI reduces the cost and time required to create content. | Documentary claim | [Anthropic research](https://www.anthropic.com/research/estimating-productivity-gains) | Verified |
| An agency was reportedly producing roughly 5,000 shows per day or week. | Uncertain recollection in transcript | [Social media report](https://x.com/sarthakgh/status/2079189926168674446) | Verified via social media report |
| AI influencers can build audiences and receive paid brand partnerships. | Documentary claim | [ScienceDirect](https://www.sciencedirect.com/science/article/pii/S219985312500126X) | Verified |
| AI-generated images have become harder to identify. | Documentary observation | [arXiv 2507.18640](https://arxiv.org/html/2507.18640v1) | Verified |
| AI-generated voices have become more realistic. | Interview observation | [QMUL research](https://www.qmul.ac.uk/news/latest-news/2025/science-and-engineering/se/ai-generated-voices-now-indistinguishable-from-real-human-voices.html) | Verified |
| Language models are effective at summarizing consensus and common patterns. | Interview interpretation | [arXiv 2507.05123](https://arxiv.org/html/2507.05123v1) | Verified |
| Widespread use of similar models can make content converge around similar frameworks and conclusions. | Documentary hypothesis | — | Hypothesis |
| Future AI systems may increasingly encounter AI-generated material in their training sources. | Documentary hypothesis | — | Hypothesis |
| Copies learning from copies may reduce originality or information quality. | Documentary hypothesis | — | Hypothesis; model-collapse research recommended |
| The internet has always had a trust problem. | Historical interpretation | — | Interpretive claim |
| People previously often assumed online content had a human creator behind it. | Documentary interpretation | — | Interpretive claim |
| Determining whether reviews, podcasts, articles, images, and creators are authentic is becoming more difficult. | Documentary claim | [NPR](https://www.npr.org/sections/money/2023/03/07/1160721021/why-we-usually-cant-tell-when-a-review-is-fake) | Verified |
| Online-only interaction can reduce empathy and awareness of other people's multidimensionality. | Interview opinion | — | Opinion |
| When information becomes abundant, credibility can become comparatively scarce and valuable. | Documentary thesis | — | Interpretive claim |
| The internet is becoming an information ecosystem that increasingly learns from itself. | Documentary thesis | — | Framing claim |