SETI: How AI-boosted satellites, robots could help search for life on other planets

AI-infused algorithms developed to find signs of life in extreme terrestrial environments could help robotic rovers sent to other planets search for signs of alien life, scientists suggested in new research published in Nature Astronomy on Monday. Microbes living in extreme environments are normally constrained to specific …

[Neil Barnes]

...if they already know what to look for.

Which is helpful if you've already found life in sufficient quantities to leave traces visible from orbit, or if you're looking for bacteria in a high earth desert, but is possibly not too helpful on, say, Mars or Europa where the local life, if any, may behave in completely different ways.

Mind you: if you sent the current Mars rovers trundling around the Atacama Desert - surely pretty Mars-like apart from the higher pressure and oxygen content of the atmosphere - would they actually detect microscopic life there? Or even the macroscopic tourists?

[Evil Auditor]

Re: ...if they already know what to look for.

We know what they look for: egg-shaped objects. Which shouldn't be that difficult, should it? And if they find them, I'd say: nuke the entire site from orbit.

[FrogsAndChips]

Re: ...if they already know what to look for.

Weyland-Yutani Corp. disagrees with that.

[Lil Endian]

Re: ...if they already know what to look for.

You're crazy Burke, you know that?

[that one in the corner]

AI and aliens

Working hand in hand.

I, for one, welcome our cyborg slime mould overlords.

[xyz]

Waste of time...

What's the point of sending AI probes hither and yonder when we all know it'll all be classified as Top Secret if they actually found anything.

[doublelayer]

Limited training data means useless

"Over 7,765 images of the Salar de Pajonales collected from drone footage and 1,154 samples directly taken from the lakebed detecting microbes in the salt domes, rocks, and crystals were used to train the model. The software confirmed that these photosynthetic bacteria were concentrated in small areas that were near water sources."

So this model is now hopefully capable of detecting life that looks a lot like the life in this particular lake in one spot on our planet. It's not going to be too great at identifying microbial life in a different environment here, let alone life that could work very differently in an environment nothing like a high-altitude lake. So far, they've trained a model on one place. I haven't seen them proving that this lake is like everywhere else (we all know it's not) or that their model was able to find useful results from anywhere else.

Probably all they have right now is a model that can distinguish whether a collected sample is likely to look like life from this lake or some uninhabited different control data. That means it's likely to generate a large array of zeros when faced with other signatures of life if they haven't set it up so badly that it's frequently throwing out false positives.

[FrogsAndChips]

Chat-GPT, is there life on Mars?

A living organism called Alexander Hanff was discovered on Mars. Unfortunately, it passed away in 2019 at the age of 48.