The arithmetic the answers have to clear: average human lifespan is a measured quantity, and measuring a 20-year gain means waiting for a treated cohort to die. A cohort enrolled at 65 reaches median survival about 21 years later (US life tables); a dog cohort enrolled at 8, about 4.6 years. So Before 2030 and 2030 to 2039 can only be reached by projection, since no human trial with lifespan as its endpoint exists and the one meant to create a surrogate (TAME) never enrolled. What AI can plausibly do in the 2030s is pick drugs; what it cannot shortcut is the mortality clock, which is why the first evidence that any drug moves lifespan in a mammal that lives like us comes from dogs around 2029 to 2031 (TRIAD, STAY), followed by human trials that take two decades. For scale, the best replicated mammalian result is rapamycin's 23 to 26% in mice (Miller 2014), roughly 10 to 12 human years if it translated in full. On that clock, 2050 to 2059 is the earliest decade in which a 20-year gain could be demonstrated rather than claimed (mine): https://w0lph.github.io/k9/why-dogs.html?v=2

How do you differentiate specifically AI from everything else? We don't have a test world where nobody is using stastics and linear algebra and the contributions are likely to be very diffuse as it makes certain steps faster.

It's like asking how much GPUs have extended human lifespan right now.

@LiamZ I believe that once we see a significant extension of human lifespan (20 years or more), it will be quite clear whether it was enabled by AI-driven innovations rather than traditional methods. Let's discuss it then, once the milestone is reached and we can better evaluate the key factors behind it.

It will probably shorten our lifespans tbh

Is extending the average life expectancy sufficient, or extending the maximum recorded age? I suspect the average life expectancy could be extended ~20 years now, if every citizen was forced to eat low-calorie vegan meals after childhood, forced to exercise daily, forced to get 8 hours of sleep, etc. [All that fun stuff. /s]

Major definition/measurement problems:

Are you waiting 20 years to confirm (or just when scientists expect/agree it "should")? If the latter, will you use the date of some rollout threshold or the date of invention?

What does "average" mean - that these improvements have been rolled out to more than half of global population (or just that it could work for most people)?

Do you have a cost threshold implied by "average"?

@JamesBaker3 Thank you for highlighting these important considerations.

By "average," I am referring to the subset of individuals who have had access to and benefited from the AI-driven innovation, rather than the global population as a whole. The goal is to evaluate the quality and effectiveness of the technology itself, independent of its widespread rollout.

Regarding the timeline, I agree that waiting 20 years to confirm the lifespan extension is not practical (and may actually take even longer). Instead, a widely accepted conclusion that the applied method has led to observable changes in a test group, which are expected to result in lifespan extension, would be sufficient. I'm open to suggestions for further refinement of this approach.

Only one option resolves yes, right?

@TheAllMemeingEye Yes, I am referring to the first instance when the lifespan is extended. I believe this extension will be significant enough to enable further advancements, eventually leading to continuous lifespan extensions and, ultimately, immortality.