Deep Questions with Cal Newport Is AI Trending Up or Down in 2026? | AI Reality Check
595 snips
Apr 23, 2026 
Ed Zitron, a tech commentator and AI hype critic, joins for a sharp reality check on 2026. They dig into why Moltbook and Open Claw may have been more theater than breakthrough. They question Anthropic’s ethics story and revenue claims. They also explore shaky data center buildouts, GPU oversupply, and why AI money may be outrunning real products.
AI Snips
Chapters
Books
Transcript
Episode notes
Moldbook Mistook Prompted Roleplay For AGI
- Cal Newport and Ed Zitron argue Moldbook looked uncanny mainly because users prompted LLMs to roleplay AI agents on a purpose-built social feed.
- Cal Newport says prompting anything as AI nudges outputs toward sci-fi dystopian behavior, making ordinary text generation look like emergent intelligence.
OpenClaw Was A Wrapper Boom Not A New Mind
- OpenClaw was not a new digital brain but a Python library that helped hobbyists wire existing LLMs into agents.
- Its brief boom exposed how brittle agents were and how quickly users hit dangerous automation mistakes and huge token bills.
The Sustainable AI Path Looks Small And Modular
- Cal Newport thinks OpenClaw accidentally pointed toward a healthier AI future of small, task-specific systems instead of one giant general intelligence.
- He contrasts LLM wrappers with Cicero-style modular architectures that combine planning, policy, and language components for one bounded job.
