Ryugu experienced heterogenous aqueous alteration and that this affected its organic materials, including some of the building blocks of life. Furthermore, the possibility that Ryugu amino acids are racemic, questions the role of astrophysical environments in the determination of single enantiomers being utilized by life on earth.
I don't usually do AI-generated comments, but this is a good case for GPT-4 explanations. GPT-4 ELI5 Ryugu experienced heterogenous aqueous alteration: Ryugu is an asteroid, which we've studied...
I don't usually do AI-generated comments, but this is a good case for GPT-4 explanations.
GPT-4 ELI5
Ryugu experienced heterogenous aqueous alteration: Ryugu is an asteroid, which we've studied using missions like Hayabusa2. This phrase means that water in some form interacted with the minerals on Ryugu in different ways across different parts of the asteroid. This interaction between water and rock, which we call "aqueous alteration," can cause changes in the rock, sometimes creating new minerals or other compounds. When we say this process is "heterogeneous," it means it happened unevenly, or in different ways in different places on the asteroid.
This affected its organic materials, including some of the building blocks of life: Organic materials are simply compounds that contain carbon, and some of these are necessary for life as we know it, like amino acids. The process of aqueous alteration can change these organic materials, possibly making them more complex or breaking them down.
The possibility that Ryugu amino acids are racemic: Amino acids, the building blocks of proteins, can exist in two forms that are mirror images of each other, like left-handed and right-handed gloves. These forms are called "enantiomers." In a racemic mix, there's an equal amount of left-handed (L-enantiomers) and right-handed (D-enantiomers) amino acids. On Earth, life almost exclusively uses left-handed amino acids.
Questions the role of astrophysical environments in the determination of single enantiomers being utilized by life on earth: This is saying that if Ryugu (an example of an astrophysical environment) has a racemic mix of amino acids, it might make us wonder how life on Earth ended up using only one type. If space environments have equal amounts of both, why does life on Earth prefer left-handed amino acids? This raises questions about how, where, and why this preference might have originated.
Everything sounds coherent, but how do I know any of this is true? I don't think AI should be used for fact checking or furthering knowledge — at least not yet.
Everything sounds coherent, but how do I know any of this is true? I don't think AI should be used for fact checking or furthering knowledge — at least not yet.
For what it's worth without sources, I used to be a planetary geologist and the summary is accurate. I think GPT was fed the article and asked for a summary. But you're right that care should be...
For what it's worth without sources, I used to be a planetary geologist and the summary is accurate. I think GPT was fed the article and asked for a summary. But you're right that care should be taken requesting such information, even summaries, from LLMs without verification.
This one is really a ball pitched right over home plate for a chatbot insofar it is a generic task (define terms which are straightforwardly defined in multiple places like “aqueous alteration” or...
This one is really a ball pitched right over home plate for a chatbot insofar it is a generic task (define terms which are straightforwardly defined in multiple places like “aqueous alteration” or “racemic”) and not something that involves specific facts, or actual reasoning (problematic in abstract domains such as arithmetic for a system which is built to get correct results, see Godel’s Theorem…. but current chatbots aren’t structurally adequate at all for reasoning instead they short-circuit along the high probability path of “conventional wisdom”, a term which was originally intended to be an insult and not a complement.). For summarization and paraphrasing though they do really well.. I think it scored 100% this time.
I prompted it with ELI5: followed by that sentence. The way to fact check it IMO is to try to understand what it’s saying and figure out if it makes sense. I don’t use it to further knowledge, I...
I prompted it with ELI5: followed by that sentence.
The way to fact check it IMO is to try to understand what it’s saying and figure out if it makes sense. I don’t use it to further knowledge, I use it to speed up the learning process.
Well okay, but someone who isn't familiar with the topic can't just do that. They can try to do that, but there's plenty of stuff in science that's true that isn't intuitive, and there are plenty...
The way to fact check it IMO is to try to understand what it’s saying and figure out if it makes sense.
Well okay, but someone who isn't familiar with the topic can't just do that. They can try to do that, but there's plenty of stuff in science that's true that isn't intuitive, and there are plenty of things that are intuitive yet aren't true. Almost all pop science falls into the latter category, and ChatGPT has certainly been trained on a lot of pop science.
Assuming that people who aren't familiar with a complex scientific topic are equipped to fact check summaries of it is how you get anti-vaxxers responding to scientists debunking their claims with stuff like "Use your own brain, would you listen to scientists if they said the sky was green?" The reality is that a lot of science just isn't something you can reason about accurately without having a background in the topic at hand, and someone who needs ChatGPT to do an ELI5 of an article about a topic is almost certainly not someone with a background in that topic.
Skepticism can be healthy, and you shouldn't blindly trust everything you read, but how do you know anything written on the web is true? I did a few quick web searches to check on terminology I...
... how do I know any of this is true?
Skepticism can be healthy, and you shouldn't blindly trust everything you read, but how do you know anything written on the web is true?
I did a few quick web searches to check on terminology I wasn't confident in my knowledge of, and I can confirm that these explanations are aligned with general knowledge available on the web. (Given gpt-4 is trained on vast amounts of text from the web, this is unsurprising.)
In chemistry, a racemic mixture or racemate (/reɪˈsiːmeɪt, rə-, ˈræsɪmeɪt/),[1] is one that has equal amounts of left- and right-handed enantiomers of a chiral molecule or salt. Racemic mixtures are rare in nature, but many compounds are produced industrially as racemates.
In chemistry, an enantiomer (/ɪˈnænti.əmər, ɛ-, -oʊ-/[1] ih-NAN-tee-ə-mər; from Ancient Greek ἐνάντιος (enántios) 'opposite', and μέρος (méros) 'part') – also called optical isomer,[2] antipode,[3] or optical antipode[4] – is one of two stereoisomers that are non-superposable onto their own mirror image. Enantiomers are much like one's right and left hands, when looking at the same face, they cannot be superposed onto each other.[5] No amount of reorientation in three spatial dimensions will allow the four unique groups on the chiral carbon (see Chirality (chemistry)) to line up exactly. The number of stereoisomers a molecule has can be determined by the number of chiral carbons it has. Stereoisomers include both enantiomers and diastereomers.
In some bodies, temperatures were modest but high enough for liquid water to exist; reaction of the original minerals with water—aqueous alteration—transformed them to complex mixtures of minerals.
I think there's an issue adding another step between the primary source and yourself -- it's like a game of telephone. There's the primary source, then the tertiary source like the encyclopedias...
I think there's an issue adding another step between the primary source and yourself -- it's like a game of telephone. There's the primary source, then the tertiary source like the encyclopedias you linked (which already should be fact checked for accuracy), and now this whole other source, AI.
There's countless examples of AI summaries being just close enough to fool most readers, but inaccurate enough to mislead them. When AIs can't find answers, they can tend to hallucinate an answer they think you'd like, and is tough to know when that's happened (especially if you're being introduced to a topic for the first time).
Also, when I'm looking for info myself, I can judge the source and how accurate I think it could be -- but not when it's an AI summary. What if it scraped "almond-farmers.com" when I asked it about the environmental impacts of almond farming?
What if a human with a bias toward almond farming answered the same question? LLMs shouldn’t really change the epistemological stance toward what you read on the web, IMO. At this point, they’re...
What if it scraped "almond-farmers.com" when I asked it about the environmental impacts of almond farming?
What if a human with a bias toward almond farming answered the same question? LLMs shouldn’t really change the epistemological stance toward what you read on the web, IMO. At this point, they’re in such widespread use that you can’t really know what’s been written by a human and what hasn’t. You should be just as critical in your thinking about anything you encounter.
I absolutely agree, but it's harder to be critical when the source is so obscured. It's not perfect when the author is a human, but when a name or account is tied to an opinion, you can at least...
You should be just as critical in your thinking about anything you encounter.
I absolutely agree, but it's harder to be critical when the source is so obscured. It's not perfect when the author is a human, but when a name or account is tied to an opinion, you can at least look at comment history or other affiliations and get a general idea where they're coming from (not perfect, but it's a step up).
At this point, they’re in such widespread use that you can’t really know what’s been written by a human and what hasn’t.
It's true and I really dislike that, which is why I don't think human posters should be copy-pasting AI answers as well.
A few years back I was thinking about the problem of sending. a factory to an asteroid like Ryugu that builds a larger factory that builds solar sails that sail back to the Earth-Sun L1 point....
A few years back I was thinking about the problem of sending. a factory to an asteroid like Ryugu that builds a larger factory that builds solar sails that sail back to the Earth-Sun L1 point. It’s a problem a bit like building a ship inside of a bottle except you are inside the bottle and trying to build a larger structure outside the bottle. Such a system would have a “stone line” that produces Silicon and Aluminum, a “metals line” that produces Iron, as well as a system that processes carbon compounds and has outputs similar to a petrochemical factory. I was taking a class on geoengineering at the time and noted that a lot of the chemistry being developed for carbon capture and utilization is relevant for asteroid exploitation, in particular you are going to get CO2 as a product of certain chemical reactions and of course you are going to recycle it all because it is precious.
The analysis published so far isn’t quite what I want for resource prospecting, but I am glad to see a lot of nitrogen compounds are there.
Two problems that still have me scratching my head are: (1) How do you capture volatile compounds? My guess is a lot of gas is going to get released when you disturb the surface, ideally you want to capture all the gas at the very beginning but you have to do that when your factory is not built out, before you’ve been able to manufacture lots of storage tanks and (2) do you send humans? If you do you’ve got the difficult problem of getting them there and keeping them alive. If you don’t you have to be able to fix anything that goes wrong remotely with many minutes of time lag.
Note that it’s been known for a long time that many asteroids are carbon rich, it’s not a completely crazy opinion that the Earth could have incorporated some hydrocarbons when it was formed, see
It's abiotic organic matter.
What does this mean?
I don't usually do AI-generated comments, but this is a good case for GPT-4 explanations.
GPT-4 ELI5
Ryugu experienced heterogenous aqueous alteration: Ryugu is an asteroid, which we've studied using missions like Hayabusa2. This phrase means that water in some form interacted with the minerals on Ryugu in different ways across different parts of the asteroid. This interaction between water and rock, which we call "aqueous alteration," can cause changes in the rock, sometimes creating new minerals or other compounds. When we say this process is "heterogeneous," it means it happened unevenly, or in different ways in different places on the asteroid.
This affected its organic materials, including some of the building blocks of life: Organic materials are simply compounds that contain carbon, and some of these are necessary for life as we know it, like amino acids. The process of aqueous alteration can change these organic materials, possibly making them more complex or breaking them down.
The possibility that Ryugu amino acids are racemic: Amino acids, the building blocks of proteins, can exist in two forms that are mirror images of each other, like left-handed and right-handed gloves. These forms are called "enantiomers." In a racemic mix, there's an equal amount of left-handed (L-enantiomers) and right-handed (D-enantiomers) amino acids. On Earth, life almost exclusively uses left-handed amino acids.
Questions the role of astrophysical environments in the determination of single enantiomers being utilized by life on earth: This is saying that if Ryugu (an example of an astrophysical environment) has a racemic mix of amino acids, it might make us wonder how life on Earth ended up using only one type. If space environments have equal amounts of both, why does life on Earth prefer left-handed amino acids? This raises questions about how, where, and why this preference might have originated.
Everything sounds coherent, but how do I know any of this is true? I don't think AI should be used for fact checking or furthering knowledge — at least not yet.
For what it's worth without sources, I used to be a planetary geologist and the summary is accurate. I think GPT was fed the article and asked for a summary. But you're right that care should be taken requesting such information, even summaries, from LLMs without verification.
This one is really a ball pitched right over home plate for a chatbot insofar it is a generic task (define terms which are straightforwardly defined in multiple places like “aqueous alteration” or “racemic”) and not something that involves specific facts, or actual reasoning (problematic in abstract domains such as arithmetic for a system which is built to get correct results, see Godel’s Theorem…. but current chatbots aren’t structurally adequate at all for reasoning instead they short-circuit along the high probability path of “conventional wisdom”, a term which was originally intended to be an insult and not a complement.). For summarization and paraphrasing though they do really well.. I think it scored 100% this time.
I prompted it with ELI5: followed by that sentence.
The way to fact check it IMO is to try to understand what it’s saying and figure out if it makes sense. I don’t use it to further knowledge, I use it to speed up the learning process.
Well okay, but someone who isn't familiar with the topic can't just do that. They can try to do that, but there's plenty of stuff in science that's true that isn't intuitive, and there are plenty of things that are intuitive yet aren't true. Almost all pop science falls into the latter category, and ChatGPT has certainly been trained on a lot of pop science.
Assuming that people who aren't familiar with a complex scientific topic are equipped to fact check summaries of it is how you get anti-vaxxers responding to scientists debunking their claims with stuff like "Use your own brain, would you listen to scientists if they said the sky was green?" The reality is that a lot of science just isn't something you can reason about accurately without having a background in the topic at hand, and someone who needs ChatGPT to do an ELI5 of an article about a topic is almost certainly not someone with a background in that topic.
Skepticism can be healthy, and you shouldn't blindly trust everything you read, but how do you know anything written on the web is true?
I did a few quick web searches to check on terminology I wasn't confident in my knowledge of, and I can confirm that these explanations are aligned with general knowledge available on the web. (Given gpt-4 is trained on vast amounts of text from the web, this is unsurprising.)
Racemic mixture
enantiomer
aqueous alteration
I think there's an issue adding another step between the primary source and yourself -- it's like a game of telephone. There's the primary source, then the tertiary source like the encyclopedias you linked (which already should be fact checked for accuracy), and now this whole other source, AI.
There's countless examples of AI summaries being just close enough to fool most readers, but inaccurate enough to mislead them. When AIs can't find answers, they can tend to hallucinate an answer they think you'd like, and is tough to know when that's happened (especially if you're being introduced to a topic for the first time).
Also, when I'm looking for info myself, I can judge the source and how accurate I think it could be -- but not when it's an AI summary. What if it scraped "almond-farmers.com" when I asked it about the environmental impacts of almond farming?
What if a human with a bias toward almond farming answered the same question? LLMs shouldn’t really change the epistemological stance toward what you read on the web, IMO. At this point, they’re in such widespread use that you can’t really know what’s been written by a human and what hasn’t. You should be just as critical in your thinking about anything you encounter.
I absolutely agree, but it's harder to be critical when the source is so obscured. It's not perfect when the author is a human, but when a name or account is tied to an opinion, you can at least look at comment history or other affiliations and get a general idea where they're coming from (not perfect, but it's a step up).
It's true and I really dislike that, which is why I don't think human posters should be copy-pasting AI answers as well.
A few years back I was thinking about the problem of sending. a factory to an asteroid like Ryugu that builds a larger factory that builds solar sails that sail back to the Earth-Sun L1 point. It’s a problem a bit like building a ship inside of a bottle except you are inside the bottle and trying to build a larger structure outside the bottle. Such a system would have a “stone line” that produces Silicon and Aluminum, a “metals line” that produces Iron, as well as a system that processes carbon compounds and has outputs similar to a petrochemical factory. I was taking a class on geoengineering at the time and noted that a lot of the chemistry being developed for carbon capture and utilization is relevant for asteroid exploitation, in particular you are going to get CO2 as a product of certain chemical reactions and of course you are going to recycle it all because it is precious.
The analysis published so far isn’t quite what I want for resource prospecting, but I am glad to see a lot of nitrogen compounds are there.
Two problems that still have me scratching my head are: (1) How do you capture volatile compounds? My guess is a lot of gas is going to get released when you disturb the surface, ideally you want to capture all the gas at the very beginning but you have to do that when your factory is not built out, before you’ve been able to manufacture lots of storage tanks and (2) do you send humans? If you do you’ve got the difficult problem of getting them there and keeping them alive. If you don’t you have to be able to fix anything that goes wrong remotely with many minutes of time lag.
Note that it’s been known for a long time that many asteroids are carbon rich, it’s not a completely crazy opinion that the Earth could have incorporated some hydrocarbons when it was formed, see
https://en.wikipedia.org/wiki/Thomas_Gold