To: US Govt, major govts, Microsoft, Apple, NVIDIA, Alphabet, Amazon, Meta, Tesla, Citi, Tencent, IBM, & 10,000+ more recipients…
From: Dr Alan D. Thompson <LifeArchitect.ai>
Sent: 28/May/2025
Subject: The Memo - AI that matters, as it happens, in plain English
AGI: 94%
ASI: 0/50 (no expected movement until post-AGI)
Author Karen Hao reporting on Dr Ilya Sutskever (15/May/2025):
Sutskever, [appeared to be] more excited and afraid than ever before of what was to come. That day, during the meeting with the new researchers, he laid out a plan… ‘We’re definitely going to build a bunker before we release AGI’…
Such a powerful technology would surely become an object of intense desire for governments globally. The core scientists working on the technology would need to be protected. ‘Of course, it’s going to be optional whether you want to get into the bunker.’ Two other sources I spoke with confirmed that Sutskever commonly mentioned such a bunker. ‘There is a group of people—Ilya being one of them—who believe that building AGI will bring about a rapture. Literally, a rapture.’
Today marks the fifth anniversary of GPT-3, the start of all this post-2020 AI stuff (paper submitted 28/May/2020). GPT-3 was the engine behind my Leta AI experiments (Episode 25, 1/Oct/2021), wrote an article for The Guardian (8/Sept/2020), and I appeared for the ABC a few years ago to explore its history and celebrate its second anniversary (May/2022).
The winner of The Who Moved My Cheese? AI Awards! for May/2025 is Elton John.
This edition has an ambient soundtrack via Udio.
We recently hit 600 large language model highlights documented on the Models Table.
Contents
The BIG Stuff (Microsoft Discovery, AlphaEvolve, StarGate UAE, Google Veo 3…)
The Interesting Stuff (AGIBOT Nezha wheels, io, Kurzweil humanoids…)
Policy (US AI regulation ban, Fair Use, Canada, DOGE issues, AI detector issues…)
Toys to Play With (WTHAPD, beyond Attention, AI stories, Grok prompts…)
Flashback (Chesterton’s fence and light…)
Next (Roundtable…)
The BIG Stuff
Microsoft Discovery AI invents a novel, non-PFAS coolant (19/May/2025)

My ASI checklist continues to document the visible markers confirming humanity’s progression into the singularity and superintelligence.
This month, Microsoft managed to partially check my ASI items #6 (first new discovery), #7 (first new physical invention), #8 (first new element added to the periodic table), and #9 (novel computing materials developed). Microsoft said:
Over the past months, we have made significant strides aiding computational scientists in their research and incorporating cutting-edge innovations from Microsoft Research. This has led to remarkable breakthroughs, such as discovering a novel solid-state electrolyte candidate that uses 70% less lithium… and enabling rapid computational simulations that accelerate scientific discoveries at Unilever…
[In May/2025], by leveraging advanced AI models… for simulation that will be available on Microsoft Discovery, Microsoft researchers discovered a novel, non-PFAS, immersion datacenter coolant prototype…
After the digital discovery process, we successfully synthesized this coolant prototype…
[Additionally, GSK, formerly GlaxoSmithKline, uses internal platforms and Microsoft Discovery] for parallel prediction and testing, creating new medicines with greater speed and precision, and potentially transforming medicinal chemistry to new unimaginable levels… [Also], Estée Lauder… is excited to harness the power of Microsoft Discovery to further accelerate the development of products…
Read the announce by Microsoft.
See it on the ASI checklist.
Watch the demonstration video, including working coolant (timecode):
In mid-2025, I think it’s likely that we are already living through the early stages of the singularity (wiki), with new inventions and discoveries already being made by large language model systems in labs around the world.
From 2025, it is reasonable to assume that novel solutions, inventions, concepts, materials, products, media, businesses, and ‘things’ are in fact being conceptualized by agentic large language model systems, even if the role of AI is deliberately obscured, and these breakthroughs are attributed to humans alone.
Prof Max Tegmark wrote about this in his 2017 book, Life 3.0. The story extract below is fiction, intended to be a prediction of a possible future with artificial superintelligence:
What the world did notice, however, was an astonishing tech boom. Upstart companies around the world were launching revolutionary new products in almost all areas. A South Korean startup launched a new battery that stored twice as much energy as your laptop battery in half the mass, and could be charged in under a minute. A Finnish firm released a cheap solar panel with twice the efficiency of the best competitors. A German company announced a new type of mass-producible wire that was superconducting at room temperature, revolutionizing the energy sector. A Boston-based biotech group announced a Phase II clinical trial of what they claimed was the first effective, side-effect-free weight-loss drug, while rumors suggested that an Indian outfit was already selling something similar on the black market. A California company countered with a Phase II trial of a blockbuster cancer drug, which caused the body’s immune system to identify and attack cells with any of the most common cancerous mutations.
Examples just kept on coming, triggering talk of a new golden age for science. Last but not least, robotics companies were cropping up like mushrooms all around the world… they dramatically disrupted the economy, and over the years to come, they gradually replaced most of the workers in manufacturing, transportation, warehousing, retail, construction, mining, agriculture, forestry and fishing.
What the world didn’t notice, thanks to the hard work of a crack team of lawyers, was that all these firms were controlled, through a series of intermediaries, by the [ASI company]. [The ASI model] was flooding the world’s patent offices with sensational inventions via various proxies, and these inventions gradually led to domination in all areas of technology.
Read more: https://lifearchitect.ai/agi-achieved-internally/
AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms (14/May/2025)
AlphaEvolve is an evolutionary coding agent powered by Google Gemini 2.0. Unlike domain-specific predecessors like AlphaFold or AlphaTensor, AlphaEvolve is designed as a general-purpose system capable of operating across a broad range of scientific and engineering tasks. It automatically modifies code and optimizes for multiple objectives, utilizing an architecture that programmatically evaluates code. This reduces reliance on human input and mitigates LLM risks such as hallucinations.
The system has demonstrated enhanced efficiency in Google's data centers, chip design, and AI training processes, and has made significant advances in solving complex mathematical problems.
Read more via Google DeepMind.
AlphaEvolve brings the total ‘alpha system’ count to 18, documented exclusively here in The Memo below, including announcement dates and links to each respective wiki page.
AlphaGo (Oct/2015)
AlphaGo Zero (Oct/2017)
AlphaZero (Dec/2017)
AlphaFold 1 (Dec/2018)
AlphaStar (Jan/2019)
AlphaFold 2 (Nov/2020)
AlphaCode (Feb/2022)
AlphaTensor (Oct/2022)
AlphaDev (Jun/2023)
AlphaCode 2 (Dec/2023)
AlphaGeometry (Jan/2024)
AlphaProof (Jul/2024)
AlphaGeometry 2 (Jul/2024)
AlphaFold 3 (May/2024)
AlphaProteo (Sep/2024)
AlphaChip (Sep/2024)
AlphaQubit (Nov/2024)
AlphaEvolve (May/2025)
Also at: https://lifearchitect.ai/gemini-report/#alpha
Google Veo 3 (20/May/2025)
Google has launched Veo 3, a cutting-edge AI video generator that uniquely integrates audio into its video creations, distinguishing it from competitors like OpenAI's Sora. Available to US subscribers of Google's Ultra subscription plan for US$249.99 per month, Veo 3 allows users to incorporate dialogue and animal sounds, enhancing the realism of AI-generated content. This launch is part of a broader suite of new tools, including Imagen 4 and Flow, aimed at expanding Google's capabilities in the generative AI space.
Read more via CNBC.
Announce: https://deepmind.google/models/veo/
Try it (paid, login): https://labs.google/flow/about
Watch some of the first generations by creators (link):
The video above includes two of my favourite clips, original sources are:
Prompt: ‘Isaac Newton rapping about gravity’ https://x.com/RubenEVillegas/status/1924917107696832569
Prompt: ‘a man doing stand up comedy in a small venue tells a joke (include the joke in the dialogue)’ https://x.com/fofrai/status/1924924738494669011?s=46
The Memo features in recent AI papers by Microsoft and Apple, has been discussed on Joe Rogan’s podcast, and a trusted source says it is used by top brass at the White House. Across over 100 editions, The Memo continues to be the #1 AI advisory, informing 10,000+ full subscribers including RAND, Google, and Meta AI. Full subscribers have complete access to all 20+ AI analysis items in this edition!
Exclusive: OpenAI’s 2025 internal strategy document (Dec/2024, released May/2025)