Ongoing important Science news thread

[photon]

I'm not sure if the article is wrong in its interpretation, or if it's first paragraph is just imprecise enough that it suggested something completely different than reality to me.

When you say the core of the Earth has stopped spinning and is spinning in the opposite direction I envision relative to itself. Which seems impossible given the level of angular momentum involved.

Then I read this article: https://www.nature.com/articles/d41586-023-00167-1

Later studies refined estimates of the rate of that ‘super-rotation’, to conclude that the inner core rotates faster than the mantle by about one-tenth of a degree per year.

The subrotation would be the core is spinning slower than the mantle, not in the opposite direction that it used to be relative to itself.

So the change would be measured in tenths of a degree per decade, which seems much more plausible

[Canada 02]

No worries, once Superman reverses time to save Lois he will set the planet back on its proper rotation

[Puppet Guy]

Doomsday Clock has been reset to 90 seconds to midnight.

WASHINGTON, D.C. – January 24, 2023 –The Doomsday Clock was set at 90 seconds to midnight, due largely but not exclusively to Russia’s invasion of Ukraine and the increased risk of nuclear escalation. The new Clock time was also influenced by continuing threats posed by the climate crisis and the breakdown of global norms and institutions needed to mitigate risks associated with advancing technologies and biological threats such as COVID-19.

Rachel Bronson, PhD, president and CEO, Bulletin of the Atomic Scientists, said: “We are living in a time of unprecedented danger, and the Doomsday Clock time reflects that reality. 90 seconds to midnight is the closest the Clock has ever been set to midnight, and it’s a decision our experts do not take lightly. The US government, its NATO allies and Ukraine have a multitude of channels for dialogue; we urge leaders to explore all of them to their fullest ability to turn back the Clock.”

https://thebulletin.org/2023/01/pres...s-to-midnight/

[puckedoff]

https://twitter.com/user/status/1631543565305405443

Not sure if this belongs here or in the "I don't wanna live on this planet anymore" thread. A further interesting development of AI application.

[pseudoreality]

Originally Posted by puckedoff

https://twitter.com/user/status/1631543565305405443

Not sure if this belongs here or in the "I don't wanna live on this planet anymore" thread. A further interesting development of AI application.

So we are building machines that can read our minds, great. There is no way that could end bad.

[TorqueDog]

You guys watch too many bloody movies.

This is huge. The use-cases for the medical field could be massive. Being able to know if someone in a coma is in pain or discomfort, or being able to communicate with someone who has had a significant injury that has left them unable to communicate through normal means.

[Fuzz]

Originally Posted by TorqueDog

You guys watch too many bloody movies.

This is huge. The use-cases for the medical field could be massive. Being able to know if someone in a coma is in pain or discomfort, or being able to communicate with someone who has had a significant injury that has left them unable to communicate through normal means.

I've read it, and don't quite understand. I need a layman's translation. I can't quite figure out if the model is entirely creating the images from the brain scan, or if it has those images in the training set, and using the keywords in reconstructing an already known image, kind of matching to what it already knows. So without the training data matching, you couldn't scan someones brain and have an image extracted of what they are thinking about. The wording is pretty impenetrable.

[TorqueDog]

Read through some of the Twitter threads from the linked comment above, it goes over some of this. Long and short of it if I'm understanding it correctly is that yes, the images were part of a massive training set (just shy of 25,000 different images) but just under 1,000 images were reserved for performing the test. But once you're at this stage, the next step is getting it to start piecing together imagery extrapolated from what imagery it has been trained on and small deviations in the fMRI scans. An example might be the patient imagining a brown teddy bear and the model picking up on that small difference and being able to render it, though it was never specifically trained on an image of a brown teddy bear. Then the next time, it might be a brown teddy bear wearing a hat instead of a ribbon or bowtie.

[Geraldsh]

Good but still just baby steps. AI will have to know everything I might know in order to picture what I know. I have seen things that nobody knows I have seen and I don’t believe AI will ever guess.

[photon]

Originally Posted by Fuzz

I can't quite figure out if the model is entirely creating the images from the brain scan, or if it has those images in the training set, and using the keywords in reconstructing an already known image, kind of matching to what it already knows.

From what I can tell the images are in the training set, so in the same way if I type "red bear" into Stable Diffusion it's going to create an image of a red bear.

What it sounds like in this case is rather than words "red bear" the input is the fMRI data.

[Fuzz]

I was a bit confused at how they mention the keywords in the paper, but I couldn't figure out where they come into play.

[Bigtime]

Sadly he qualifies this as "not aliens":

https://twitter.com/user/status/1636013347224133635

Still will be interesting to see what has been discovered.

[Ozy_Flame]

Originally Posted by pseudoreality

So we are building machines that can read our minds, great. There is no way that could end bad.

Time to watch Minority Report

[Muta]

Originally Posted by Bigtime

Sadly he qualifies this as "not aliens":

https://twitter.com/user/status/1636013347224133635

Still will be interesting to see what has been discovered.

Liquid water? That's my guess

[OutOfTheCube]

Originally Posted by Bigtime

Sadly he qualifies this as "not aliens":

https://twitter.com/user/status/1636013347224133635

Still will be interesting to see what has been discovered.

These things always turn out to be exciting for science people but boring for regular people. It'll be like, oh there's evidence of a volcano erupting 500 years ago, or we found some weird particles in the air that could be a sign of microscopic life, which is cool but doesn't really mean anything unless you're invested in science stuff.

Now if it's like oh we found some alien ruins or discovered some full sized alien bugs or animals, that would be some ####.

[Ozy_Flame]

It's just going to be granular scientific milestones that mean absolutely nothing to most people. Same thing that is announced regularly through organizations like NASA, academia, etc. Probably the discovery of some chemistry in the crust or atmosphere. Don't get your hopes up for anything mind-blowing.

[Erick Estrada]

Originally Posted by TorqueDog

You guys watch too many bloody movies.

This is huge. The use-cases for the medical field could be massive. Being able to know if someone in a coma is in pain or discomfort, or being able to communicate with someone who has had a significant injury that has left them unable to communicate through normal means.

You should play Atomic Heart as it may change your perception of AI involvement with a human's mind.

[Jiggy_12]

Originally Posted by OutOfTheCube

These things always turn out to be exciting for science people but boring for regular people. It'll be like, oh there's evidence of a volcano erupting 500 years ago, or we found some weird particles in the air that could be a sign of microscopic life, which is cool but doesn't really mean anything unless you're invested in science stuff.

Now if it's like oh we found some alien ruins or discovered some full sized alien bugs or animals, that would be some ####.

Haha, nailed it. It’s an active volcano

https://twitter.com/user/status/1636066117344014343

[Knut]

I really should not have put money on it being The Proto-Molecule.

[surferguy]

Originally Posted by Knut

I really should not have put money on it being The Proto-Molecule.

It certainly would make things more interesting