slashdave

Wait... and people use this?

No myth.

Software is utterly trivial compared to the life sciences. The reason is quite simple: reproducibility and explainability.

Do you have any idea of the amount of money involved in pharmaceutical R&D?

> as long as you have a suitable robot to run a lab

Let me guess. You have never stepped in a lab.

> we're going to see massive increases in parallelization

Why? This is not an AI problem, it's an industry one. And if it was simple, it would already exist.

Keep waiting

Your outrage is well earned, but fairly rare. The reason nothing is done is because the public doesn't care.

Unlocking is the exact opposite of security.

What problem are you trying to "solve"?

Hard to find an easier thing to build...

Not necessarily. It is often due to a lack of choices (in lead candidates).

There would be less failures if problematic therapies were filtered out before the clinical stage.

I guess you prefer this: https://en.wikipedia.org/wiki/United_States_Triumphal_Arch

This is not a marketing problem, many ML scientists have also fallen into this trap.

> other three copies of your iMessage data on Apple’s servers are just, uh, not encrypted

Because they want to allow people to recover them.

> It's ostensibly for security, but it's more about control.

The security angle is easy to understand.

Not sure why Apple cares about control.

But it easier just to want to make Apple seem bad, so there's that.

Very few

Yeah, most people cannot be bothered with (3) honestly

Yup. And a real market: "hip" kids buying such a device for their parents.

You do realize that those two goals conflict.

or perform a tail vein injection or oral gavage

Sounds like you have never worked in a lab.

Um, it's called peer review. Dumping random crap as preprints isn't progress.

Good luck trying to judge experimental papers.

This is backwards. Not sure why what is essentially a PR effort by these companies automatically becomes an advancement in the field.

Do we want these capabilities to be useful? It seems making them accessible so that the existing community can use it as a tool is the right approach. Automated proof methods are already used like that.

> Make informed decisions regarding your use of these products

They are mass marketed. The creators should do the upmost to ensure this and not pin blame on users.

> There'd also be fewer people falling for apocalyptic Rationalist delusions.

Um, this is coming from the researchers building these models. So, good luck?

Liability is enough of a reason

> based on perceived values of the expected outcome

Not true either

Like any technology. Make it a crime when appropriate, otherwise expose bad behavior to legal liability.