Concepts

AI's Impact on Society: Benefits and Challenges

Published by When Notes Fly

https://whennotesfly.com/concepts/systems-complexity/what-will-ai-do-to-society

When Notes Fly article, fact-checked and corrected.

Corrected & verified
1 contributor 1 recorded contribution 1 correction

Who created and checked this page

Correction

Emir BaycanExpert Contributor

Correction history

Every accepted correction to this page is recorded with the exact change, so readers can see how the page improved over time.

  1. 28 July 2026 · corrected by Emir Baycan

    Qualified seven overstated or dated AI claims: reframed LLM 'emergent reasoning' as actively debated (possible pattern-matching / measurement artifact) rather than an established capability; qualified 'the majority of large employers use screening tools' and the Amazon automatic-termination claim (described as reporting, with managerial review noted); corrected AlphaFold from 'virtually all known proteins' to over 200 million computational structure predictions of varying confidence; attributed the '300 million jobs' figure to Goldman Sachs (2023) and clarified it measures task exposure, not net job loss; attributed the frontier-compute doubling estimate to sources like Epoch AI and flagged the 6-9-month figure as sample- and period-dependent; and noted that the October 2023 U.S. AI Executive Order was revoked in January 2025 and is no longer a current framework. The Acemoglu-Restrepo robot-employment figure (0.2-0.3 workers per robot per thousand) was verified as accurate and left in place.

    Before

    The article stated that LLMs develop emergent reasoning at scale, that a majority of employers use screening tools and Amazon automatically terminates workers, that AlphaFold predicted virtually all known protein structures, that generative AI could automate 300 million jobs, a 6-9-month compute doubling since 2012, and presented the 2023 AI executive order as a current framework.

    After

    Qualified the emergent-reasoning, screening, and Amazon claims; corrected the AlphaFold scope; attributed and scoped the 300-million-jobs and compute-doubling estimates; and updated the executive-order status to reflect its January 2025 revocation.

    Why: These claims were more certain, precise, or current than the evidence supports. Corrections attribute figures to their sources (Goldman Sachs 2023; Epoch AI; DeepMind AlphaFold) and update time-sensitive policy facts rather than adding new sources.

    View the full record →

Full contribution timeline

  1. Correction 28 July 2026

    Emir Baycan: Qualified seven overstated or dated AI claims: reframed LLM 'emergent reasoning' as actively debated (possible pattern-matching / measurement artifact) rather than an established capability; qualified 'the majority of large employers use screening tools' and the Amazon automatic-termination claim (described as reporting, with managerial review noted); corrected AlphaFold from 'virtually all known proteins' to over 200 million computational structure predictions of varying confidence; attributed the '300 million jobs' figure to Goldman Sachs (2023) and clarified it measures task exposure, not net job loss; attributed the frontier-compute doubling estimate to sources like Epoch AI and flagged the 6-9-month figure as sample- and period-dependent; and noted that the October 2023 U.S. AI Executive Order was revoked in January 2025 and is no longer a current framework. The Acemoglu-Restrepo robot-employment figure (0.2-0.3 workers per robot per thousand) was verified as accurate and left in place.

    Proof record →