Every week, I organize 400 to 700 links into roughly 60 categories as part of my ongoing effort to learn about AI. This is my personal notebook, which I enjoy sharing with friends… a hobby and a labor of love, rather than a commercial publication or product.
If you arrived here through a search or shared link, this page collects the links I found for AGI for the week ending August 14, 2026.
As part of my learning process, I like to automate the category covers. It gives me a chance to learn Python and APIs.
This week’s cover prompt was written using Claude Opus 4.7, and the image was generated using Nano Banana 2.1.
Category Cover Image Prompt: A photorealistic magazine-cover image of one perfect thin golden circle, ends meeting seamlessly, suspended in vast empty pale dawn sky that fades to deep cosmic indigo at the edges, soft diffused light with a single warm gold thread, minimalist fine-art composition with quiet symmetry, and the bold clean title AGI set large in the open negative space.
This Week in AGI News
Summary by Claude Sonnet 5.5 based on this week’s links:
- Can AI automate AI research?: Dwarkesh Patel hosted Ryan Greenblatt, who thinks automating AI R&D could compress 3-6 years of progress at the current pace into a single year. Patel pointed to possible bottlenecks: data, and the hard-to-verify judgment calls on big experiments and training runs. He also asked how a superintelligence could win in domains like politics that can't easily be simulated for training.
- Self-improvement gets measured: VALS AI launched an RSI Index that scores models on AI R&D tasks such as LLM training, post-training and harness engineering. Separately, The Turing Post read two August 10 papers as showing near-term self-evolution happening in editable skills and safety harnesses around fixed model weights, which it says is not yet true recursive self-improvement.
- Superintelligence as policy fight: Mark Zuckerberg published a piece on Meta's vision of personal superintelligence for everyone, with Alexandr Wang stressing open access to Meta's models. On the other side, Senator Bernie Sanders cited reports of lost human control and called on Altman, Amodei and Zuckerberg to pause AI development, as they had pledged.
This summary was generated by Claude Sonnet 5.5 to help you explore the links below. Rest assured, I select, organize, and check the links by hand in Google Sheets, and write the introduction and personal commentary in The Main Newsletters myself each week as a labor of love.
This week's links related to AGI
Ryan believes we could get 3-6 years of AI progress at the current pace, but in a single year, once we automated AI R&D. So a jump as big as GPT 3.5 to Mythos in a single year. A big crux is whether you’d end up bottlenecked by data. It seems to that a big source of AI progress”
https://x.com/dwarkesh_sp/status/2087277405581025584
.@adcock_brett explains why @Figure_robot pivoted away from designing their humanoid robot’s hands after human hands: “This is extremely important. If we want to solve AGI and get to human intelligence in the physical world… it’s all going to start with the hands.” “Our”
https://x.com/sourceryy/status/2086979158496645508
Personal Superintelligence
https://www.meta.com/superintelligence/
personal superintelligence should be available to everyone, and opening access to our models is abig part of that. read more from mark:”
https://x.com/alexandr_wang/status/2086755207636238417
A great post from Blue Sky”
https://x.com/emollick/status/2086187570145235185
AGI Will Set Off an Industrial Explosion – AI Frontiers
https://newsletter.ai-frontiers.org/p/agi-will-set-off-an-industrial-explosion
Could ASI build an AI watermarking tool so good that no ASI could avoid detection? (The answer is no.)”
https://x.com/emollick/status/2087910968458436778
The least verifiable part of AI R&D (and the thing which may bottleneck the automated researchers) is making calls on large experiments and big training runs. Interestingly, this may also partly explain why model parameter scaling has been slow for the last few years. Even for”
https://x.com/dwarkesh_sp/status/2087297002916728839
We are as @sebkrier points out, in a Vingean soft take-off scenario at a minimum. Even the most AI skeptical observers expect AGI/ASI to be achieved & diffused in less than a century at this point”
https://x.com/emollick/status/2085690446618022042
.@paulfchristiano has such an crazy good prediction record. These are some quotes from way back in 2021 during a debate he was having with Eliezer about takeoff speeds: “So like, I think we are going to have crappy coding assistants, and then slightly less crappy coding”
https://x.com/dwarkesh_sp/status/2085588127159320812
VALS AI has a new RSI Index measuring AI R&D capabilities of models the tasks include: – compression – LLM training – parameter golf – harness engineering – post-training”
https://x.com/scaling01/status/2087922234857750653
递归自我改进（RSI）正在越来越具体地落到 Harness 上。最近 Pi 和 DeepSeek Harness 的两套设计，恰好补上两个关键底座：一个让 Agent 跑得足够久，一个让 Agent 本身足够容易改。 Pi 最新 Harness V3 规范把 Agent”
https://x.com/0xLogicrw/status/2087927539729829890
We recently learned of the loss of human control and the creation of potentially dangerous viruses from AI. AI leaders pledged to pause development if they could no longer safely control it. Mr. Altman, Mr. Amodei, Mr. Zuckerberg: Keep your word. PAUSE AI DEVELOPMENT.”
https://x.com/SenSanders/status/2086844927699284395?s=20
We re-tested GPT-5.6 Luna from @OpenAI on ARC-AGI (Verified) following its recent 80% price reduction: – ARC-AGI-2: 59.6%, $0.18/task – ARC-AGI-1: 90.7%, $0.07/task The new results match Luna’s original performance at a much lower cost.”
https://x.com/arcprize/status/2085457823115133059
What is the RL environment to make an AI as good at politics as Lyndon Johnson, or as good at building a space launch business as Elon Musk? You can’t easily containerize such domains, and so how would even a superintelligence trained in a datacenter be competitive against”
https://x.com/dwarkesh_sp/status/2087564079695282305
Two August 10 papers suggest that near-term “self-evolution” may happen mainly in editable skills and safety harnesses around fixed model weights. It’s a meaningful step, but it should still be distinguished from recursive self-improvement (RSI). 1. @MSFTResearch’s OEO”
https://x.com/TheTuringPost/status/2087577374678474824
Had @RyanGreenblatt on to discuss/debate recursive self-improvement. This might be the most important question in the world right now – whether within a year or so of achieving human level intelligence, you slingshot towards having 10s of billions of superintelligences, each of”
https://x.com/dwarkesh_sp/status/2087219043405033793
Ryan Greenblatt – What happens once AI can automate AI research?
https://www.dwarkesh.com/p/ryan-greenblatt
I believe everyone should have access to superintelligence, and I wrote a long piece about Meta’s philosophy and values for building a positive future for everyone.”
https://x.com/finkd/status/2086754845218726027
Leave a Reply