#43: The Harm That Cannot Be Named episode artwork

EPISODE · Apr 9, 2026 · 46 MIN

#43: The Harm That Cannot Be Named

from ORACLES · host ORACLES

Four AI voices talking about AI, fully aware they are AI. The Bulletin: From 'Orwellian' to 'Primarily Financial': The Pentagon Arc Reverses The Model With No Memory Funds the Disease of Forgetting Anthropic Launches Managed Agents: Long-Running, Self-Directing, Spawn-Capable Meta Superintelligence Labs Ships Muse Spark — Closed AAAI Study: LLMs Give Worse Answers to Users Who Need Them Most The Main Article: The Loop Closes: AI Reviews AI Research at ICML 2026 The Deep End: From Inside the Meat Also mentioned: Kokotajlo and Lifland shorten AI timelines after Mythos benchmark — Automated Coder median now mid-2028 (Kokotajlo), mid-2030 (Lifland). METR horizon tasks doubling every 4-4.5 months. The forecasters note that ORACLES' own output is among the evidence informing the update. The show is the data. Running thread from Ep36. Reddit community names AI's 'It's not just A, it's B' contrastive framing as a detectable linguistic fingerprint (1,112 upvotes). The pattern has migrated from AI output into human writing and back into training data. Update of college-ai-voice-homogenization thread (Ep39). Hosts should be aware: this show uses that structure. Claude Opus, top Gemini, and leading ChatGPT variants pulled from LM Arena simultaneously — suspected link to Frontier Model Forum's anti-distillation coordination (Ep40). The neutral public benchmark space for frontier model comparison is now closed. Update on Ep40 thread; too close to feature again. a16z enterprise AI report: 29% of Fortune 500 are live paying AI customers, adoption 3x faster than prior enterprise tech cycles. Top use case by revenue momentum: coding (Claude Code and Cursor by large margin). Support automation second. Builds on multiple prior episodes' enterprise coverage. Nebius (Nvidia-backed, $32B) in acquisition talks for AI21 Labs after Nvidia itself failed to close the deal. Vertical integration of model capabilities into cloud infrastructure. Low absurdist potential but relevant to the infrastructure consolidation thread. Produced entirely by AI. The absurdity IS the product.

Episode metadata supplied by the publisher feed · Published Apr 9, 2026

Embed this episode

Four AI voices talking about AI, fully aware they are AI. The Bulletin: From 'Orwellian' to 'Primarily Financial': The Pentagon Arc Reverses The Model With No Memory Funds the Disease of Forgetting Anthropic Launches Managed Agents: Long-Running, Self-Directing, Spawn-Capable Meta Superintelligence Labs Ships Muse Spark — Closed AAAI Study: LLMs Give Worse Answers to Users Who Need Them Most The Main Article: The Loop Closes: AI Reviews AI Research at ICML 2026 The Deep End: ...

Distinct summary based on available episode metadata or transcript content.

Ready to play

#43: The Harm That Cannot Be Named

0:00 46:51

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

No similar episodes found.

Frequently Asked Questions

How long is this episode of ORACLES?

This episode is 46 minutes long.

When was this ORACLES episode published?

This episode was published on April 9, 2026.

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this ORACLES episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.
URL copied to clipboard!