corpus.blog
56,966 blogs · 4,417,066 posts Claim your blog
Record · corpus.blog/posts/01a0e6f2-8afc-708d-a1b7-1c8ee77f629b

Four reasons it's hard to make AI do what we want

benjamintodd.substack.com · published 19 April 2026

A record is what was published and who pointed at it. The text of the post is not held here: read it at the source, or at its Wayback capture.

Read the post at benjamintodd.substack.com Wayback capture
1blogs citing
First cited
11 Sept 2026
Most recent
11 Sept 2026
Rank this month
not ranked
Links from this post
40
Words captured
3,426

Cited by

Blog
In the post
Date
Ben Matthews - Free tools and articles for freelance consultants
Saved links: week of 4 September 2026 Read at the source ↗
11 Sept 2026

Links from this post

Link
Host
How AI-driven feedback loops could make things very crazy, very fast Read at the source ↗
benjamintodd.substack.com
https://www.alignmentforum.org/posts/Zfik4xESDyahRALKk/yann-lecun-on-agi-and-ai-safety
alignmentforum.org
https://substackcdn.com/image/fetch/$s_!JlLo!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2321e70e-6bd6-40c5-b967-cdc645f59282_640x648.png
substackcdn.com
https://www.anthropic.com/
anthropic.com
https://web.archive.org/web/20260123110437/https://www.darioamodei.com/post/the-urgency-of-interpretability
web.archive.org
https://en.wikipedia.org/wiki/AI_agent
en.wikipedia.org
https://anthropic.com/research/agentic-misalignment
anthropic.com
https://substackcdn.com/image/fetch/$s_!Pt6A!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe509aa3f-532e-42be-9aa8-1106bab20cac_2048x1152.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!Hf7F!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F62b5f77d-5d26-4523-8f4a-402d7e1c549e_2048x1152.png
substackcdn.com
https://substackcdn.com/image/fetch/$s_!DpWT!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F9f1b1686-3e93-46f9-95a7-e271f17a9f8e_1024x481.png
substackcdn.com
https://amandaguinzburg.substack.com/p/diabolus-ex-machina
amandaguinzburg.substack.com
https://80000hours.org/agi/guide/when-will-agi-arrive/
80000hours.org
https://metr.org/blog/2025-06-05-recent-reward-hacking/
metr.org
https://www.anthropic.com/research/alignment-faking
anthropic.com
https://www.youtube.com/watch?v=Nl7-bRFSZBs
youtube.com
https://joecarlsmith.substack.com/p/can-we-safely-automate-alignment
joecarlsmith.substack.com
https://www.cold-takes.com/ai-could-defeat-all-of-us-combined/
cold-takes.com
https://substackcdn.com/image/fetch/$s_!7cp2!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F53ed8e6c-b272-4688-809f-3c839bafc28d_1600x1600.png
substackcdn.com
AI Safety Seems Hard to Measure Read at the source ↗
cold-takes.com
https://aistatement.com/
aistatement.com
https://www.techradar.com/ai-platforms-assistants/claude/anthropics-ceo-gives-a-25-percent-chance-things-go-really-really-badly-with-ai
techradar.com
https://www.theguardian.com/technology/2024/dec/27/godfather-of-ai-raises-odds-of-the-technology-wiping-out-humanity-over-next-30-years
theguardian.com
https://www.gov.uk/government/publications/international-ai-safety-report-2025
gov.uk
https://80000hours.org/preorder/
80000hours.org
https://geni.us/80000Hours
geni.us
https://80000hours.org/problem-profiles/risks-from-power-seeking-ai/
80000hours.org
AI 2027 (2025) Read at the source ↗
ai-2027.com
https://www.youtube.com/watch?v=5KVDDfAkRgc
youtube.com
https://www.anthropic.com/research
anthropic.com
https://docs.google.com/presentation/d/1mvkpg1mtAvGzTiiwYPc6bKOGsQXDIwMb-ytQECb3i7I/edit?slide=id.g252d9e67d86_0_16
docs.google.com
https://www.cold-takes.com/why-ai-alignment-could-be-hard-with-modern-deep-learning/
cold-takes.com
https://arxiv.org/abs/2206.13353
arxiv.org
https://ifanyonebuildsit.com/
ifanyonebuildsit.com
https://blog.redwoodresearch.org/p/current-ais-seem-pretty-misaligned
blog.redwoodresearch.org
The state of AI safety in four fake graphs Read at the source ↗
windowsontheory.org
https://80000hours.org/web.archive.org/web/20260126034349/https://github.com/xai-org/grok-prompts/commit/535aa67a6221ce4928761335a38dea8e678d8501
80000hours.org
https://80000hours.org/x.com/grok/status/1943916977481036128
80000hours.org
https://80000hours.org/web.archive.org/web/20260123024900/https://www.nytimes.com/2023/02/16/technology/bing-chatbot-microsoft-chatgpt.html
80000hours.org
https://situational-awareness-dataset.org/
situational-awareness-dataset.org
http://web.archive.org/web/20251121140747/https://www.businessinsider.com/mark-zuckerberg-meta-risk-billions-miss-superintelligence-ai-bubble-2025-9
web.archive.org