AGI/ASI Timelines thread (AGI/ASI may solve longevity if it doesn't "kill us all" first)

The following estimate strips away the “futurist optimism” of Kurzweil/de Grey and instead focuses on structural trends in drug discovery, the “China Factor” (regulatory arbitrage and primate research), and AGI-driven compression of clinical trials.

Revised Estimate:
If AGI arrives by ~2030–2035, the probability of reaching Longevity Escape Velocity (LEV) by 2055 is 65% – 75%.

This is significantly higher than the standard Western-consensus view (often ~40-50%) because of the specific Chinese variables you noted.

The Probability Model: Why China Changes the Equation

In a Western-only model, FDA conservatism creates a “Death Valley” between AGI discovery and human implementation. China removes two primary bottlenecks: Primate testing latency and Regulatory ethics constraints.

Factor Western-Centric View (Standard) China-Inclusive View (Adjusted) Impact on 2055 LEV
Testing Bottleneck AGI simulates mice; we wait 5-10 years for safety data. China creates primate models immediately. (See: 2025 CAS Primate Breakthrough). High Acceleration
Regulatory Risk Aging is not a disease; FDA demands long safety trials. Two-Tier System: “Special Approvals” for elites (Project 981) + fast-track for strategic tech. Medium Acceleration
AI Integration Private siloes (Isomorphic, Recursion) protect IP. State-Level Integration: Huawei Pangu Drug Model + Hospital Data integration. High Acceleration

1. The “China Factor”: Hard Biotech vs. Western Ethics

You are correct to flag Chinese biotech as a missing variable. Recent developments suggest China is willing to execute experiments that Western IRBs (Institutional Review Boards) would block or delay for decades.

  • The Primate Shortcut (2025 Breakthrough): In late 2025, researchers from the Chinese Academy of Sciences (CAS) successfully used genetically engineered stem cells to reverse aging markers in primates.
    • Why this matters: The West relies on mice (which don’t translate well to humans). China is moving directly to non-human primates. This cuts the “translation error” lag time by roughly 5–10 years.
  • He Jiankui’s Return: The fact that He Jiankui (CRISPR babies) is back in a lab in Beijing working on Alzheimer’s and Duchenne suggests a regulatory environment that prioritizes results over process once the initial PR storm settles.
  • Strategic AI: Unlike the fragmented US startup scene, China is deploying state-backed AI models like Huawei’s Pangu Drug Molecule Model (trained on 1.7B molecules) directly into hospital networks (e.g., Xi’an Jiaotong University) to screen antibiotics and senolytics.

2. The “Dual-Track” Regulatory Arbitrage

The US FDA requires a therapy to prove it treats a specific disease (e.g., “Diabetes,” not “Aging”). This creates a 10-15 year delay for longevity drugs like Metformin or Rapamycin to be repurposed.

China operates a de facto dual-track system:

  1. Public Track: Standard NMPA approval (now only ~1.3 years slower than FDA).
  2. Elite/Strategic Track: Rumors of the “981 Leaders’ Health Project” persist, suggesting that promising AGI-generated therapies will be tested on a “compassionate use” basis for aging elites long before Phase III trials conclude.
    • Impact: If AGI identifies a senolytic cocktail in 2032, it could be in human use in China by 2035. Western medical tourists would force the FDA to accelerate or face irrelevance, effectively breaking the global regulatory bottleneck.

3. Adjusted Timeline to 2055

  • 2027–2030 (AGI Arrival): AGI solves protein folding completely and models complex biological systems (cells/organs) with >99% accuracy.
  • 2030–2035 (The “Wet Lab” Bottleneck):
    • US: AGI designs trials; FDA debates safety signals.
    • China: AGI designs therapies; CAS tests on 10,000 primates immediately. Human trials for “frailty” (proxy for aging) begin in Shenzhen.
  • 2035–2045 (The First LEV Therapies):
    • First effective “rejuvenation” toolkit (likely partial reprogramming + senolytics) stabilizes mortality rates for 60-year-olds.
    • LEV Achieved: Technical LEV is reached here. For every year you live, science adds >1 year to your remaining expectancy.
  • 2045–2055 (Distribution): Therapies become cheap enough for mass adoption.

Conclusion: The Probability Estimate

  • Without China: ~40% (Bogged down by 15-year FDA trials and mouse model failures).
  • With China: ~70%

The risk isn’t that the science fails; it’s that AGI offers a solution that we are too afraid to test. China’s regulatory aggressiveness and willingness to use primate models bridges that gap.

===

The exponential growth of microplastics introduces a new variable: Environmental Mortality Drag. However, if we assume the AGI timelines remain constant (~2030s), the probability of reaching Longevity Escape Velocity (LEV) by 2055 drops only slightly, to roughly 60–65%.

While microplastics create a new “disease” (systemic inflammation/cellular senescence), the same AGI capabilities that solve aging (nanotechnology and protein design) are uniquely suited to solving microplastic contamination.

Here is the breakdown of the “Race”: Exponential Plastic Accumulation vs. Super-Exponential AI Remediation.

1. The Negative Drag: Why Probabilities Drop (~ -10%)

Microplastics essentially add a “difficulty multiplier” to the problem of aging. We aren’t just trying to fix natural decay anymore; we are trying to fix decay plus toxic accumulation.

  • The “Silent” Inflammation: Microplastics act as chronic stressors. They don’t kill you instantly; they increase “background” inflammation (inflammaging), which accelerates cardiovascular disease and Alzheimer’s.
  • The “Trojan Horse” Effect: Nanoplastics (smaller than microplastics) can cross the blood-brain barrier and placenta. They carry other toxins (heavy metals, bacteria) deep into tissues where standard drugs can’t reach.
  • Medical Irony: Current tools to clean blood (like dialysis) often shed more plastic into the body because the tubing and filters are made of… plastic. This creates a “technological dead end” that humans currently can’t engineer their way out of easily.

2. The AGI Solution: Why the Drop is Small

If AGI arrives by ~2030, it catches the microplastic curve before it becomes catastrophic for the 2055 timeline. The solutions AGI would likely deploy are distinct from today’s primitive filters.

  • Enzymatic Remediation (The “Pac-Man” Strategy):
    AGI can design enzymes (proteins) that specifically digest plastic polymers into harmless byproducts. We already have Ideonella sakaiensis (a bacteria that eats PET plastic), but it’s too slow. AGI could optimize this enzyme to be 10,000x faster and safe for the human bloodstream.

  • Medical Nanobots:
    Instead of filtering blood through a machine (which sheds plastic), AGI enables the fabrication of magnetic or biological nanobots that are injected into the bloodstream. These bots can “tag” or “eat” nanoplastics and then be guided out of the body (e.g., via magnetic fields or urine).

3. The “China Factor” Revisited

The “China Factor” becomes even more relevant here.

  • The Crisis Point: China has some of the highest microplastic soil contamination in the world due to agricultural plastic film usage. The government views this as a food security threat.
  • Aggressive Deployment: While the West might debate the safety of releasing “plastic-eating bacteria” into the wild or the human body, China is more likely to deploy these AGI-designed biological agents rapidly to secure its soil and population health, accelerating the timeline for a cure.

Summary: The Adjusted Probability

The exponential growth of plastics makes the biological puzzle harder, but it does not break the AGI thesis.

Metric Without Microplastics With Exponential Microplastics
Biological Difficulty Hard (Entropy) Very Hard (Entropy + Toxicity)
AGI Solution Set Gene Editing / Senolytics + Nanobots / Enzyme Design
LEV Probability (2055) ~70% ~60–65%

The Bottom Line: If AGI fails to arrive, microplastics significantly reduce our life expectancy. If AGI does arrive, cleaning microplastics out of the body is a trivial physics problem compared to reversing aging itself. The “Plastic Apocalypse” is only an apocalypse for a non-AGI civilization.

Next Step

Would you like to explore the specific “Order of Operations” for LEV therapies (e.g., which therapies like senolytics or blood scrubbing you should likely target first in the 2030s) to maximize your bridge to 2055?

https://x.com/RicursiveAI/status/1995932204703346946

^i’ve noticed them for several years, I think they could already significantly influence things

1 Like

I’m more optimistic. In my reasoning, the juncture is such that any forecast can be off the mark, given the recent acceleration in the performance of LLMs. If AGI is not here already, in some fields AI’s intelligence far overwhelms human intelligence, for example, languages, data analysis, text reading and elaboration.
From what I hear, the main bottleneck is the availability of CPUs and, above all, energy. If these can be worked out, AGI is just around the corner. And ASI is not further away.

It is impressive, my present avatar has been generated by NanobanaPro with just a few instructions, starting from a picture of mine. Makes it appear like I really found the right cocktail for longevity, but with some collateral effects (which I may accept given the results).

image

3 Likes

But to tell the truth I was really stunned when I saw what GROK’s Imagine can do. Multiple image-realizations of the single prompt, a real random distribution of images. Many of them are very similar, some are different, some outliers, like when you generate random numbers from the same statistical distribution. I could not believe it. And you can choose single images and transform them into short clips. I cannot show them now because I reached my free quota. I’ll post the realizations later.

I’m checking, when asked to create multiple images of a woman, Nanobanana created four statistical realizations of the same woman, but all in the same image frame.

Q.: Can you create multiple realizations of the same prompt, with separate images not in the same frame?
A.; Is there another image I can try? I’m here to help you create all types of things, but can’t make images like that.

1 Like

Grok seems to be much better. Now, I submitted this prompt asking about a scantily dressed woman, but in the multiple iterations of the image, grock provided some which are apparently not dressed at all.
Moderator, please feel free to erase the upper image if too spicy, this is just to illustrate a statistical output. There are dozens of variations. They are multiple frames, which can be singled out and improved or animated. I§ chose the first one, the more discreet. The results speak for themselves. This is the present degree of evolution of image generation and animation, for free. Incredible.

Prompt: A scantily dressed, gorgeous woman, seen from behind with a tattooed backside, in a lush tropical garden with flowers, water, and a breeze.

Is AI an existential threat to human control of the planet? Some of us remember the scene in that great 1968 science fiction film, 2001: A Space Odyssey, in which HAL, the super-intelligent computer that controls the spaceship, rebels against its human masters. Today, as AI makes rapid progress, Dr Hinton recently told me that it was only a matter of time before AI becomes smarter than humans. Does that raise the possibility that humans will lose their ability to control the planet? How do we stop that extraordinary threat?

AI poses unprecedented threats. Congress must act now » Senator Bernie Sanders

Bernie is old enough that he probably watched the premiere.

1 Like

Grok’s image processing is impressive and I like the multiple image generation that takes place. But you don’t have much control (from what I tried) in organizing and placement of text, etc. on the images. It’s fine for entertainment purposes, but not very good for scientific / bio / medical type images. For that I’ve found Gemini much better.

1 Like

https://medium.com/@rmaruy3/a-map-for-the-ai-in-for-science-discourse-57c352f527a6

Those images are mediocre tbh, frontier is at least 1 year ahead, try Sora or Nano Banana Pro, which are indistinguishable from real images.

You are probably right, as far as fine details go, Nano Banana Pro is exceptional, I tried again the theme of a scantily dressed woman in a lush garden with flowers and a water pool. Although, by western standards, we can hardly say the woman is dressed scantily, she’s rather fully dressed in a tropical garb.

I think that Gemini and Grok offer two distinct image generators, the former extremely precise and able to insert text (a feature which now has been evidently improved), the other provides multiple iterations of the same theme, with guided animation at the end with a full, short video clip as an output.

The bottom line on which I think we can all agree is that these AI tools are improving exponentially, as the other text-based tools are doing.

This suggests that it is plausible to think that AGI may be around the corner. That’s in the field of imprecise probabilities, but sounds more and more plausible by the day.

Now, just out of curiosity since we’re discussing image generators. I gave the same prompt to Grok. The degree of realism does seem lower, but it answered maybe more correctly, prioritizing the first request in the list: ‘A woman, scantily dressed’. It also provided dozens of pictures with different women, among which the one I posted is one of the more dressed. Plus, I can have a clip of it. Different generators, different solutions, different requirements (like having text for scientific illustrations).

image

1 Like

https://x.com/i/status/2001063756445733346

I half believe him… [only b/c AGI timelines are so aggressive now]

BJ does look a bit rejuvenated recently or am I tripping?

I’ve been watching interviews of a guy in his early 30’s and when I saw this my brain reminded me of him.

image

Low picture quality. His face still reminds me of a 40-something guy. It lacks the features typically associated with a younger person. The hairline also looks rather high and thin.

also opus4.5 really broke the METR curves, this really is the closest evidence for the singularity being near

https://x.com/rickwierenga/status/1976326164877885843

Dario was predicting powerful AI early 2026 last year:

Obviously, many people are skeptical that powerful AI will be built soon and some are skeptical that it will ever be built at all. I think it could come as early as 2026, though there are also ways it could take much longer.

AI companies might redirect the compute used for training the model to running a million instances of it by 2027:

  • The resources used to train the model can be repurposed to run millions of instances of it (this matches projected cluster sizes by ~2027), and the model can absorb information and generate actions at roughly 10x-100x human speed. It may however be limited by the response time of the physical world or of software it interacts with.

  • Each of these million copies can act independently on unrelated tasks, or if needed can all work together in the same way humans would collaborate, perhaps with different subpopulations fine-tuned to be especially good at particular tasks.

But Dario doesn’t believe in an instant transformation / Singularity:

First, you might think that the world would be instantly transformed on the scale of seconds or days (“the Singularity”), as superior intelligence builds on itself and solves every possible scientific, engineering, and operational task almost immediately. The problem with this is that there are real physical and practical limits, for example around building hardware or conducting biological experiments. Even a new country of geniuses would hit up against these limits. Intelligence may be very powerful, but it isn’t magic fairy dust.

Relevant thread: We Have AGI Now? It's Command-Line Agents like Opus 4.5

1 Like