Peyman Shahidi on X: "(3/10) The core idea is: AI enables “chaining” of steps. Delegate a run of consecutive steps to AI; the human checks only the final output of the chain. The chain can be viewed as a single “meta-task,” in which the last step is augmented and all preceding steps are automated. https://t.co/1MnZdq3eqb" / X
To view keyboard shortcuts, press question mark View keyboard shortcuts Home Explore Notifications Chat Grok Premium Bookmarks Creator Studio Articles Profile More Post Kavish Garg @thatkavish Post See new posts Conversation Peyman Shahidi @Peyman_Shahidi · Feb 23 (1/10) New working paper with @demirermert , @johnjhorton , @immorlica , and Brendan Lucier! We study how agentic AI can reshape worker tasks and jobs. We view GenAI not just as a productivity booster, but as a tool that changes how workflows get chopped up. 12 86 329 68K Peyman Shahidi @Peyman_Shahidi · Feb 23 Link: https:// shorturl.at/Fftn8 Production is a sequence of steps. Each step can be executed in one of three modes: (1) Manual: human does it without AI’s help, (2) Augmented: AI does it and human validates the output, (3) Automated: AI does it without direct human oversight. nber.org Chaining Tasks, Redefining Work: A Theory of AI Automation 1 3 29 3.7K Peyman Shahidi @Peyman_Shahidi (3/10) The core idea is:
Peyman Shahidi @Peyman_Shahidi Replying to @Peyman_Shahidi (3/10) The core idea is: AI enables “chaining” of steps. Delegate a run of consecutive steps to AI; the human checks only the final output of the chain. The chain can be viewed as a single “meta-task,” in which the last step is augmented and all preceding steps are automated. 1:38 PM · Feb 23, 2026 3.6K Views 1 0 1 2 0 2 23 0 2 3 5 0 5 Read 1 reply
Explore this link on the map →related reading
- When AI builds itself \ Anthropicanthropic.com
- After Automation | Everyevery.to
- Current AIs seem pretty misaligned to me — LessWronglesswrong.com
- AI Horseless Carriages | koomen.devkoomen.dev
- Measuring AI Ability to Complete Long Tasks - METRmetr.org
- We spent 2 hours working in the future - METRmetr.org
- My AI Adoption Journey – Mitchell Hashimotomitchellh.com
- Convictionconviction.com
- AI and the human creativity cycle | mathu's thoughtsmathurah.com
- Current AIs seem pretty misaligned to meblog.redwoodresearch.org
- Building Effective AI Agents \ Anthropicanthropic.com
- Convictionconviction.com