The Memo by LifeArchitect.ai

The Memo by LifeArchitect.ai

Share this post

The Memo by LifeArchitect.ai
The Memo by LifeArchitect.ai
The Memo - 30/Jun/2025

The Memo - 30/Jun/2025

$1T 'Project Crystal Land' AI fab in Arizona, CEBSIT BMI, 6th Neuralink user Rob Greiner, and much more!

Dr Alan D. Thompson's avatar
Dr Alan D. Thompson
Jun 29, 2025
∙ Paid
10

Share this post

The Memo by LifeArchitect.ai
The Memo by LifeArchitect.ai
The Memo - 30/Jun/2025
Share
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:    30/Jun/2025
Subject: The Memo - AI that matters, as it happens, in plain English
AGI:     94%
ASI:     0/50 (no expected movement until post-AGI)

Prof Jeff Clune, DeepMind, former research manager OpenAI (20/Jun/2025):
I made this analogy the other day when I was talking to a journalist. He was asking me what it was like to be at OpenAI and places like that a couple of years back. The analogy I gave was: it’s kind of like you’re an astronomer, and you’re looking at your equipment, your sensors and your computer readouts and everything. And you, along with a handful of other people, have the expertise to look at this complicated data and say, ‘Oh my gosh, aliens are on the way. They’re going to arrive on Earth in a couple of years.’

We haven’t fully figured out how fast they’re traveling or what their technology is, but they’re going to be here real soon. And that’s going to change everything… will they usher in a new era of great technology, friendships, knowledge expansion? Who knows?

We seven people—or seventy people—knew it. But nobody else listened to us or believed it. I was speaking to journalists, and nobody was listening. Everyone thought we were crazy. But in that room, we knew. The data were pretty clear: we were sure the aliens were coming.

And the analogy is even weirder, because we weren’t just watching the aliens come, we were also giving them information. We were helping them come to us by making the technology. That dynamic makes it even stranger.

Another reason the analogy fits so well is this: you walk home at the end of the day, and people are pushing babies in strollers and going to the supermarket. Everything looks totally normal. It doesn’t feel like the world is about to radically change. But rationally, you know it is. You know the world is going to be wildly different very soon.

I still feel that way. Yes, now the world has woken up, and a lot more people are having these conversations. Prime ministers are aware. But you still have skeptics saying, ‘You’re just making this up, hyping it for your own valuations. It’s hitting a wall. Nothing’s about to happen.’

But I’m very confident the world is going to be radically different very soon. And I don’t believe the world deeply understands that yet, at least, not as much as it needs to.

There’s a soundtrack for this edition, Apple’s new human-made ringtone for the upcoming iOS 26 (20/Jun/2025), and the AI-made club remix by Suno AI (21/Jun/2025)!

1×
0:00
-0:10
Audio playback is not supported on your browser. Please upgrade.
1×
0:00
-2:51
Audio playback is not supported on your browser. Please upgrade.

In The Memo edition 11/Jan/2025, I included an empirical piece on brain atrophy due to LLM use:

Drawing from my background in human intelligence research and as a former chairman of Mensa International’s gifted families, the evidence suggests that widespread use of large language models (LLMs) such as ChatGPT and Claude for everyday tasks—emails, trivia, research, ‘life admin’—seems to already be fostering distinct mental shortcuts that may lower our cognitive fitness. This ‘lazy brain’ phenomenon might persist until we seamlessly merge AI into our neural pathways via brain-machine interfaces (BMIs)…

This erosion of mental capabilities through AI use will stem from reduced mental practice and diminished active engagement. While the exact magnitude is yet to be quantified, I welcome research in this area.

A few readers were miffed, some were very angry, asking why I would write about something they hadn’t heard about. (That’s exactly why The Memo exists, and is often forward-looking.) After the edition went live, the evidence on LLM contribution to brain atrophy began pouring in:

  • Jan/2025: Swiss Business School: ‘AI tools in society: Impacts on cognitive offloading and the future of critical thinking‘

  • Jan/2025: Microsoft + CMU: ‘The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort…’

  • Feb/2025: UTexas: ‘Protecting human cognition in the age of AI’

  • Jun/2025: MIT: ‘Your brain on ChatGPT: Accumulation of cognitive debt…’

This month, MIT found that ‘While LLMs offer immediate convenience, our findings highlight potential cognitive costs. Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels.’

Contents

  1. The BIG Stuff ($1T AZ fab, CEBSIT BMI, Neuralink 6 and 7, Gemini next…)

  2. The Interesting Stuff (xAI compute, AlphaGenome, redundant LLMs, Bob McGrew…)

  3. Policy (Military LLMs, CBNR risk increasing, Fair Use, OpenAI v MSFT, Chinese safety…)

  4. Toys to Play With (Magenta realtime, Doppl, New voice assistants, Therapist AI…)

  5. Flashback (GPT-2…)

  6. Next (GPT-5 is coming, Roundtable…)

The BIG Stuff

Exclusive: Mid-2025 AI retrospective: The sky is delivering (26/Jun/2025)

Image by GPT-4o image gen. https://lifearchitect.ai/the-sky-is-delivering/

Full subscribers to The Memo received my mid-year AI report last week. The final section, ‘What’s coming for you’, outlines ten major new human connections being brought to life by AI.

Read the report: https://lifearchitect.ai/the-sky-is-delivering/

SoftBank’s Son pitches $1 trillion Arizona AI hub (20/Jun/2025)

SoftBank Group's visionary founder, Masayoshi Son, is exploring the development of a US$1 trillion industrial complex in Arizona focused on robotics and AI. This ambitious project, dubbed Project Crystal Land, seeks collaboration with Taiwan Semiconductor Manufacturing Co (TSMC) and aims to establish a high-tech manufacturing hub akin to China’s Shenzhen. With the potential to significantly bolster US tech manufacturing, SoftBank is in discussions with US federal and state officials to explore incentives for companies investing in this groundbreaking initiative.

Read more via Reuters.

Sidenote: The numbers being thrown around in 2025 are outrageous, though not surprising. Two years ago in my mid-2023 report, I noted:

Whether it’s 1 billion or 100 billion or even a trillion dollars of investment, this news [of OpenAI’s $29B valuation in 2023] is not particularly surprising. Indeed, any organization that leads us to artificial general intelligence has untold value. But, in a post-scarcity world—and perhaps in a post-capitalist world and a post-money world—that value will not be measured in dollars.

The Economist also published a piece called ‘AI valuations are verging on the unhinged: Unless superintelligence is just around the corner’ (25/Jun/2025)

CEBSIT BMI: China launches first-ever invasive brain-computer interface clinical trial — Tetraplegic patient could skillfully operate racing games after just three weeks (15/Jun/2025)

‘Now I can control the computer with my thoughts. It feels like I can move at will.’ Source: CEBSIT.

Shanghai-based Center for Excellence in Brain Science and Intelligence Technology (CEBSIT), has launched China’s first ever clinical trial of high-throughput wireless invasive BCI. On 5/Mar/2025, through minimally invasive surgery, they successfully implanted a BCI system in a patient with tetraplegia. Three weeks post-implant, 25/Mar/2025, the tetraplegic patient, who lost his limbs in an accident, successfully controlled a computer and operated racing games with his mind.

Comparing the new CEBSIT BCI with Neuralink, they say the former has “a cross-sectional area only 1/5 to 1/7 that of Neuralink's electrodes and flexibility over 100 times greater.” It adds that the neural electrodes used are the smallest and most flexible in the world… minimize damage to surrounding brain tissue and provide better prognosis over the longer term… 26mm in diameter and under 6mm thick. [Neuralink is 23mm x 8mm]

These BCI devices have now been built into China’s current legislation, ready for 1M+ users, and impacting my ASI checklist item #43: https://lifearchitect.ai/asi/

According to the state broadcaster CCTV, China's National Healthcare Security Administration on March 12 released the Guidelines for Establishing Pricing Items for Neurological Medical Services, which specifically designated separate pricing items for BCI technologies, including "Invasive BCI Implantation Fee" and "Invasive BCI Removal Fee." Once local authorities align with and implement these guidelines, BCI medical service fees will have a standardized basis. This means that, as soon as the BCI technology matures, the pathway for service fee charges in clinical applications is already in place.

Read more via Tom's Hardware, and the Global Times source.

7 people now have Elon Musk's Neuralink brain implant (27/Jun/2025)

Source: Reddit.

Neuralink has expanded its revolutionary brain-computer interface to seven patients, offering individuals with cervical spinal cord injuries or ALS the ability to control computers with their thoughts. Since February 2025, Neuralink has more than doubled its patient count and secured a US$650M funding round. The PRIME study, conducted by Barrow Neurological Institute, is testing the safety and efficacy of the N1 implant, the R1 surgical robot, and the N1 User App, marking significant strides in advancing human-computer interaction.

Read more via PCMag.

For the 31st International Symposium on Controversies in Psychiatry, I explored the R1 surgical robot, the N1 implant, and more in my opening keynote ‘Generative AI: GPT-5 and brain-machine interfacing’ (available to full subscribers of The Memo).

The Memo - Downloads + highlights (2025)

The Memo - Downloads + highlights (2025)

Dr Alan D. Thompson
·
May 28
Read full story

Interview with 6th Neuralink user Rob Greiner (Jun/2025)

Absolutely amazing what I can do already in under a week with Neuralink! I can't thank the Neuralink team, the doctors, the surgeons and all of my family and friends support. It may only look like I'm doing a couple games but it's been under a week and I'm already shocked! (Rob, 18/Jun/2025)

Watch the video (link):

The Interesting Stuff

Exclusive: Solace (26/Jun/2025)

One common theme I’ve noticed with audiences around the world is an underlying sense of fear or trepidation about AI. I’ve tried many, many different avenues to have people see the enormous benefits of having ‘more smarts’ in the world.

My latest visualization leverages research into high-ability humans, where we uncovered a significantly increased sense of ethics and moral sensitivity (Sep/2011). Frontier AI models seem to have even higher levels of these qualities (2023). The integration of AI with humans is expected to take us to the peak of intelligence and ethical refinement.

Take a look: https://lifearchitect.ai/solace/

Read my related 2023 paper, ‘Endgame’: https://lifearchitect.ai/endgame/

The 23% solution: Why running redundant LLMs is actually smart in production (9/Jun/2025)

I was going to put this in the Toys to Play With section, but I think this is important, simple, and we’re going to see a lot more of it happening under the hood.

Optimizing AI-based systems often involves creative solutions to latency issues. By running two (or more) models in parallel, the author dramatically improved response times, reducing average latency by 23.2% and nearly eliminating long waits over 10 seconds.

Instead of switching models, I now fire requests to both Gemini 2.5 Flash AND GPT-4o simultaneously, returning whichever responds first.

This strategy leverages the strengths of both models, offering redundancy, natural load balancing, and resilience against individual model outages, all while keeping costs manageable given the declining token prices.

Read more via Reddit.

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 25+ AI analysis items in this edition!

The newspaper that hired ChatGPT (13/Jun/2025)

Il Foglio, an Italian newspaper, has embraced generative AI by incorporating ChatGPT into its editorial process, producing a daily insert of AI-written articles. Editor Claudio Cerasa views AI as an unavoidable force, likening it to the wind that must be managed. The publication plans to continue using AI for its weekly section, showcasing the potential of AI as a creative tool in journalism.

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2025 LifeArchitect.ai
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture

Share