has 691 lines. I expect it would work, as FAWK seems to be a very simple language. I'm currently working on a similar project with a different language, and the equivalent AST module is around 20,000 lines and only partially implemented according to the standard. I have tried to use LLMs without any luck. I think in addition to the language size, something they currently fail at seems to be, for lack of a better description, "understanding the propagation of changes across a complex codebase where the combinatoric space of behavioral effects of any given change is massive". When I ask Claude to help in the codebase I'm working in, it starts making edits and going down paths I know are dead ends, and I end up having to spend way more time explaining why things wouldn't work to it, than if I had just implemented it myself...
We seem to be moving in the right direction, but I think absent a fundamental change in model architecture we're going to end up with models that consume gigawatts to do what a brain can do for 20 watts. Maybe a metaphorical pointer to the underlying issue, whatever it is, is that if a human sits down and works on a problem for 10 hours, they will be fundamentally closer to having solved the problem (deeper understanding of the problem space), whereas if you throw 10 hours worth of human or LLM generated context into an LLM and ask it to work on the problem, it will perform significantly worse than if it had no context, as context rot (sparse training data for the "area" of the latent space associated with the prior sequence of tokens) will degrade its performance. The exception would be like, when the prior context is documentation for how to solve the problem, in which case the LLM would perform better, but also the problem was already solved. I mention that case because I imagine it would be easy to game a benchmark that intends to test this, without actually solving the underlying problem of building a system that can dynamically create arbitrary novel representations of the world around it and use those to make predictions and solve problems.
Just tried. Actually quite impressed with how well it works. I avoid using AI to write code, I'm a little worried that the existence of detection tools like this will lead people to over-rely on them; I would feel bad if someone suggested I used AI to create code I took pride in writing. I don't matter, but on a societal scale that effect may compel people to over-rely on AI as their work is treated as slop whether they put effort in or not, which will just increase the tide of terrible AI slop code, engineers managing systems they do not understand, and thus the brittleness and instability of global infrastructure. I sincerely hope you guys succeed, I suppose the point is that almost succeeding might be worse than not trying at all...
This is the state equivalent of a human gouging out their own eyes to avoid dealing with the fact that a train is barrelling towards them instead of stepping off the tracks.
The data this satellite collects is used to promote climate regulation and legislation. The authors of the article can’t even come up with other ostensible uses for it. Farmers care about carbon dioxide levels? Good one.
The equivalent would be a democrat administration using government force to shut down data collection they don’t like because it will be used to promote policies they don’t want. Kind of like how the Biden admin’s FAA restricted drone use at the southern border during the migrant crisis.
I think the word you’re looking for is “like” instead of “equivalent”.
In the sense that restricting drone use does not destroy assets that were expensive to create and launch, and are currently generating significant value for the American taxpayer and the world. We already paid for them, and now we're maintaining them for next to nothing.
Also, note that the yearly maintainance cost is in the same order of magnitude as a single presidential golf trip. TBC, I'm not against presidential leisure, everyone needs a break sometimes - but the cost argument doesn't hold up.
Additionally, the migrant crisis is specific to current American political environment. Climate change is an issue which threatens humanity, and even if one is purely self-interested and doesn't care for famines in the third world, will directly cost the taxpayer trillions of dollars of the next century via extreme weather events and relocation/insurance bailouts, and indirectly cost hundreds of billions more via stunted productivity.
Restricting drone use for petty political reasons is not equivalent to pro-actively de-orbiting functional satellites that are working to mitigate existential risk.
Just because it cost money to put the satellite into orbit doesn’t mean that it has value. Would a buyer pay $15M a year for the data produced by this satellite? Maybe, but the fact that the government put it into orbit and not a private company is telling.
Realistically, because it is owned by the government, it’s worth exactly what the government says it is. Just ask the Biden administration who sold materials for the border wall far below cost.
How does this satellite mitigate existential risk? We already know carbon dioxide contributes to climate change. We should spend money on reducing it, not on costly projects that tell us things we already know.
Speaking of existential risk, the US national debt is a far more urgent and certain risk than climate change. At least decommissioning this satellite helps on that front. How much carbon gets removed from the atmosphere if we keep it operational?
> Just because it cost money to put the satellite into orbit doesn’t mean that it has value.
True. But the fact that it was actively generating valuable data does.
> Would a buyer pay $15M a year for the data produced by this satellite? Maybe, but the fact that the government put it into orbit and not a private company is telling.
Not really. One of of the first things you learn in an economics class is the notion of a "public good": goods which are provided by the government that increase the wellbeing of society, but for which no individual actor in society is incentivized to create or maintain. Street lights and interstate road networks are the prototypical examples, the original GPS constellation might be another candidate.
> Realistically, because it is owned by the government, it’s worth exactly what the government says it is.
EDIT: My original response to this point was weak, the revision is: certain subsets of the government (like NASA) consider it valuable. Other subsets of the government are seeking to destroy it because they don't like the data it produces. The latter subset does not consist of scientists, and has an active political interest in suppressing that information. I think that the former subset of the government is more capable of an accurate value assessment in this context.
> Just ask the Biden administration who sold materials for the border wall far below cost.
I do not have access to Biden admin officials and have no idea what you're talking about so I'll take your word for it.
> How does this satellite mitigate existential risk? We already know carbon dioxide contributes to climate change. We should spend money on reducing it, not on costly projects that tell us things we already know.
Understanding a threat is a necessary precursor to defeating it. In a concrete sense, this data is necessary for models that facilitate prediction of effects that inform policy decisions we make. Metaphorically, if you're in a forest and you know that a grizzly bear is out to get you, knowing the direction of and arrival time of the bear useful information. Except that doesn't do it justice, because in this metaphor, we've already constructed a lookout tower, and you're taking the position that burning down the lookout tower will save us the effort of having to climb the ladder, and we already know a bear is coming for us anyways so who cares?
> Speaking of existential risk, the US national debt is a far more urgent and certain risk than climate change. At least decommissioning this satellite helps on that front.
I agree that the national debt is a serious issue, but I already addressed the cost point in a previous comment.
> How much carbon gets removed from the atmosphere if we keep it operational?
Directly? None. Indirectly? Speculative. Whether it directly removes carbon is different than whether it provides insight which allow us to more efficiently and effectively deal with the risk by predicting how it will materialize, which is objective anyways. Historically, a strategy of "we're going to destroy assets that provide valuable information because we don't like the reality they present" has never worked out.
A morbidly obese person doesn’t need a scale that is accurate to thousandths of a pound. They need medical intervention.
We don’t need a satellite that tells us there is increasing carbon dioxide in the atmosphere that is warming the globe with extreme accuracy. We need to do something about it.
You are making the mistaken assumption that more data collection directly leads to less carbon dioxide in the atmosphere. The steps that need to be taken to reduce it are the same whether or not you can measure atmospheric carbon dioxide with extreme accuracy or not.
The morbidly obese person can lose weight without even taking measurements. The results will be obvious if they’re doing it right. The same is true with climate change. You’ll see fewer droughts, less hurricanes, wildfires, etc.
To use your analogy, you would rather sit in your watchtower and keep tally of exactly how many bear teeth pierced the skin of victims on the ground instead of climbing down from the tower and shooting the bear.
I respect your view and appreciate that you seem to be acting in good faith. I'm also glad to see that we agree that climate change is a serious issue that needs to be dealt with.
This is an instrument that cost hundreds of millions of dollars. More than half a billion considering the others in the program. It is still producing data. If you are arguing in favor of destroying this asset, then you have ulterior motive or terrible judgement.
The fact that it cost hundreds of millions to put these satellites into orbit is irrelevant. That’s sunk cost. The only relevant question is whether the data collected from them is worth $15M a year. I don’t think it is.
Think about all the carbon you could sequester with $15M a year. You could lease thousands of acres of farmland in the Midwest, plant trees, and sequester thousands of tons of carbon for that amount of money.
But no, we must keep spending money on projects like this to keep the climate hysteria at a maximum. It’s hard to produce headlines like “Atmospheric Carbon Rises to a New High Under Trump administration” without this satellite, you know.
Remind me, who has ulterior motives or terrible judgement?
Go talk to a farmer. They pay attention to a lot of metrics, but local carbon dioxide levels are not one of them.
Use your brain. If farmers actually found this data valuable, the author of that article would have found someone who was using it and interviewed them.
The word is just a 'pointer' to the underlying shared experience, so I don't think you could; the kid would come away thinking red is the same thing as the feeling of splinters or the warmth of a sunset, which isn't what red is, those are just feelings maybe associated with red. That said - I'm actually pretty confident we'll be able to have basic "conversations" made of basic snippets of information with dolphins and whales in my lifetime. Maybe not complex grammatical structure we identify with, but small stuff like: "I'm hungry". I'm not sure if dolphins could understand "fish or octopus for dinner?", because they might not have any idea of a logical 'OR', and perhaps they might don't even differentiate between fish/octopus.
We do share (presumably) experiences of hunger, pain, happiness, the perception of gradations of light and shape/form within them, some kind of dimensionally bound spatial medium they exist in as an individual and are moving through - though of course they might not conceive of these as "dimensions" or "space", they would surely have analogs for directional words - although given they aren't constrained to live on top of a 2D surface, these might not be "up", "down", "left", "right", but something in the vein of "lightward" or "darkward" as the two main directions, and then some complicated 3D rotational coordinate system for modeling direction. Who knows, maybe they even use quaternions!
For the subset of shared experiences and emotions this should be possible, not only that, I feel that we must try (as in, it's a moral/ ethical obligation).
Training an ML on dialogues alone will not be enough. One would need to spend a lot of time to build up a wealth of shared experiences, so that one can learn the mapping/correspondence.
