# If AI 2027 is right, are we really going to make patients wait 15 years for treatments that will soon exist? — retained record

> Retained reddit record from reddit:singularity.

Canonical: https://highsignal.app/data/records/71e4ea23c440279b

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:singularity)  published 2026-09-23 09:01 ingested 2026-09-24 02:38

[original source ↗](https://www.reddit.com/r/singularity/comments/1wo0pac/if_ai_2027_is_right_are_we_really_going_to_make/)

## retained text



```
If AI 2027 proves even remotely accurate, we could be looking at AI systems capable of dramatically accelerating biomedical research within the next few years.

That raises an obvious problem; if AI starts discovering genuinely life changing treatments at a pace far beyond anything we have seen before, how do we deal with the fact that current clinical trial and pharmaceutical regulatory structures can still take 10 to 15 years to move a treatment from discovery to widespread patient access? I work in the clinical trials space and the uptake of AI here is extremely minimal, whereas in drug discovery it is already revolutionising the space - the gap is immense.

At some point very soon the bottleneck will stop being scientific discovery and become validation, regulation and deployment. Some of that is inevitable, but not all, or even most of it. Most of those 10-15 years are not spent actively testing medications of animals or humans.

Given the pace of AI progress and assuming it continues, 15 years is a laughable amount of time. By the time a drug that slows the progression of ALS by a few months is devised by one agent reaches the market, another agent will have devised a total cure for the disease many times over.

How do we handle a world where medications highly likely to cure or transform countless life changing illnesses are delayed by processes that are completely unready for the pace at which they will arrive?
```



document fetched 2026-09-24 00:33 · canonical https://reddit.com/r/singularity/comments/1wo0pac/if_ai_2027_is_right_are_we_really_going_to_make
