Interesting thread on why it's not possible for the solar panels to be self cleaned: https://twitter.com/NASAInSight/status/1590736220199735296?c...
Amazing how in 15 years they never managed to fix this UX issue, despite have ten times more people than they actually needed.
>Amazing how in 15 years they never managed to fix this UX issue, despite have ten times more people than they actually needed.
It's not been established at all that this is the case. If they manage to keep things running without running afoul of consent decrees, etc. for a few years, then you might have a point.
Cars were invented in 1886, the first windshield wipers became available for them in 1917, 31 years later. That is...
- in an environment with plenty of water - not near waterless,
- on glass windshields (much less fragile than solar panels),
- where we can physically replace the wipers when needed,
- where we can adjust if they get stuck,
- and where there is much less dust.
If we add 5 years of R&D to overcome each of those obstacles (very optimistic, it's more like 10 well-funded years I suspect), then even optimistically we're looking at 25 years. I suspect we might have a rover with self-cleaning panels, or safe cold-fusion, in 25 years. I don't expect one this decade.
"Once you’ve designed something once it’s a lot faster to design over and over".
You forgot the last part of that sentence: Once you’ve designed something once it’s a lot faster to design over and over, if you are in a similar environment with somewhat similar conditions.
oh, now i understand the math
Someone should spend a couple years doing some PhD work on Martian dust blowing turbine designs, but not burn out completely so they can write a 1000 word, interesting blog post with some pretty pictures for me to consume.
I suspect it has something to do with the non-linear response of the human eye. You don’t have to reflect a lot of light to be perceived as significantly lighter than black, so diverting 25% of the light from the panel to reflected light makes the panel look quite light.
And little Mars-dust-wiper pads have lesser ... pads, and so ad infinitum.
It's pad's all the way down.
NASA plans obsolescence into the rovers to get funding to build the next rover.
NASA is by no means perfect. But we can start by assuming they have basic competence and positive intent unless proven otherwise.
That said, I also realized that a motorcycle helmet tear off plastic film and some rubber bands held stretched by fusible links would solve the problem just fine and at least double the life of the mission for negligible weight. It may not be glamorous or even guaranteed to be environmentally friendly, but it would work. Kind of like the old joke about NASA spending millions to design a space pen when the Soviets just used a pencil.
So I'm afraid that there's still no way I can accept that the engineering isn't feasible. It just sets off my code smell detector too much. Planned obsolescence is perhaps a bit too strong of an accusation though, so I'm sorry for saying that.
Since drones already expand the abilities of a lander, i.e. exploring the immediate area and collecting samples, this wouldn't be much of an added cost.
I guess the downside of having JPL in SoCal is that none of the engineers are familiar with windshield wipers. <jk>
I went out of it thinking, "yep, look at these excuses we are making for the state of robotics." - the quote is: "A system like that would have added cost, mass, and complexity. The simplest, most cost-effective way to meet my goals was to bring solar panels big enough to power my whole mission – which they did (and then some!)."
So robotics is at the state where it is possible to send a rover to Mars, but a simple robotic arm can't easily brush off some dirt. The lander weighs 789 lb and the cost-effective solution is not try include a simple cheap lightweight robot arm that is foolproof and can easily sweep some dirt, because there is not such thing. Robotics isn't at the level where that is easy, cheap, or lightweight.
If a human could reach through a portal to there with a brush, it would take less than five minutes to brush off the dirt that had fallen on solar cells.
A simple, light, and foolproof robotic arm that can easily do that doesn't exist. Now that we have superhuman levels of general AI in chatgpt, isn't it time to work on foolproof and lightweight robot arms for all sorts of tasks?
Given the very low atmosphereric pressure on Mars, a bit of gas at terrestrial sea level pressure equivalent would do wonders.
As for downsides - you need a compressor and some filtering to avoid clogging your gas system with dust (thats likely the main obstacle).
Still doable IMHO and I'm sure followup probes and eventually settlements will find how to handle Martian dust efficiently.
A brush doesn’t work well in a zero moisture environment.
The dust is horrifically abrasive and brushing it along the panels would cause damage.
The added weight would have to be compensated for.
If Insight dies then Spirit will have lasted 50% longer and opportunity significantly longer, but both rovers lost quite a bit of functionality over time. So, the difference isn’t as stark as you might assume when you consider how much time rovers spent with reduced power or hibernating, stuck, etc.
Another analogy might be to Uber, which has revolutionized the transportation industry by offering a new way of connecting riders with drivers through a smartphone app. Uber may not meet some traditional definitions of a "taxi company," but it has proven to be a successful and popular service for millions of people.
A third analogy might be to the smartphone itself, which has become an essential tool for many people despite not being a traditional computer. Smartphones have disrupted the personal computing industry and offer a wide range of capabilities, from making phone calls and sending texts to accessing the internet and running apps.
Similarly, ChatGPT is able to complete many general requests that meet the expectations of many users, even though it might not meet some definitions of general AI. It is able to do this because it is able to perform a wide range of tasks, including writing this comparison at your [my] request.
Of course, ChatGPT has its limitations and is not a substitute for human intelligence. However, it is able to complete many tasks that would be challenging or impossible for chatbots of decades ago, which demonstrates its usefulness and capabilities as a tool for assisting users with a wide range of tasks.
I hope this helps clarify my personal feelings about ChatGPT's intelligence and capabilities. It is important to recognize that different people have different standards and definitions for what constitutes general AI, and that is completely fine. What matters is whether a tool is able to meet the needs and expectations of its users, and ChatGPT has certainly been able to do that for me and many other users.
--
[1] I want to clarify that this response was written by ChatGPT, a large language model trained by OpenAI, at my request. For more context on my request and the interaction with ChatGPT, you can view the full exchange at https://hastebin.com/raw/wolovegiqa
In case you click through, what I want you to focus on in the transcript is that it came up with the two additional analogies at the end at my request, but I absolutely didn't suggest or introduce what they are specifically and I had no idea what it would come up with. You can't know this, but it's my experience.
Understanding and coming up with analogies is a gold standard of general intelligence, and it also understood my request, which was of a highly generic nature.
You might argue that it could have picked better analogies but I gave it pretty strict requirements to focus on technology companies. At this point we are arguing about whether airplanes really fly like birds, or whether submarines really swim. It's irrelevant. ChatGPT gets from point A to point B which is the point. It's some form of AGI. Just make it a generic request if you don't believe me, ask it to something generic that has never been done before and it'll do a great job at it. Go ahead and see for yourself.
You can make chatgpt generic requests and it answers them. just try it if you don't believe me.
I asked it how it would solve not being able to get some cookies off of a high counter if it were a small child, it replied with its plans:
https://hastebin.com/raw/geponewiki
I had it generate a PDF of the latest advances in AI. It gave me something pretty generic and not that insightful. See for yourself:
http://online.verypdf.com/app/sharepdf/?url=http://online.ve...
That is a PDF titled "AI Advances:A Review of the Latest Developments in Artificial Intelligence". Nothing special as far as titles or contents goes. But an AI wrote the whole damn thing.
(Though it couldn't fix the broken formating when I tried to make a latext document out of it.)
But that's not the point. The point is I asked it to create an executive summary of advances in AI and it did as well as a seventh or eighth grader.
What you really don't understand is you can't just instantly apply the standard of "this thing must be the best at everything and never make any mistake" to be generally intelligent. Ask it some intelligence type of tests of a generic nature.
I mean what's more generic than some new question nobody has any reason to ask and that requires inductive and deductive reasoning to solve? https://hastebin.com/raw/gupevutohe
That's the definition of AGI. I mean what is your standard? What is AGI supposed to do that this thing doesn't do at all?
I just think that this meets my requirements for handling generic arbitrary tasks. Its input and output is language but it is able to think and keep track of complex thoughts including about entirely novel situations. It's usually pretty reasonable.
What more evidence of AGI do you need than that it can solve novel generic tasks it has never encountered before?
"""GPT (Generative Pre-training Transformer) is a type of machine learning model developed by OpenAI that is used for natural language processing tasks, such as language translation, summarization, and question answering. GPT is trained on large amounts of text data and is able to generate human-like responses to prompts.
AGI, or artificial general intelligence, refers to the ability of a machine or artificial intelligence system to perform any intellectual task that a human being can. AGI is often thought of as a hypothetical future technology that would be able to understand and learn any intellectual task that a human being can, rather than being narrowly specialized to perform specific tasks like current AI systems.
GPT is not equivalent to AGI, as it is a specific type of machine learning model that is designed to perform natural language processing tasks, rather than being able to perform any intellectual task that a human being can. However, some researchers believe that the development of AGI may require the use of machine learning models like GPT as a building block, as they can be used to train AI systems to understand and generate human-like language."""
It similarly told me it is also not a "form of AGI"
There is a chatgpt detector here: https://detectchatgpt.com/
I asked ChatGPT to write a story (my prompt was just "Write a story about a pumpkin"). ChatGPT wrote a nice story, well, slightly weird, and the chatgptdetector detected it with 99.96% confidence ("We estimate a 99.96% probability that this text was generated by ChatGPT or another GPT variant.") while a visual bar showing its confidence filled up all the way.
I next gave ChatGPT the instruction "Rewrite it so it is not detected as GPT output." (thanks to a tip on Reddit, where I saw this mentioned.)
Bear in mind that ChatGPT is a GPT variant. I am asking it to fool a test that by definition it cannot fool. It would be like asking you to figure out how you can go through a human detector and not be detected as human.
I fully expected the site to identify it despite ChatGPT's best attempt at self-obfuscation since by definition it is still outputing GPT output.
This time it passed the test. The bar dropped from a full 99.96% to just 15%.
It was the same story. ChatGPT just successfully passed the generic request to fool some detection algorithm it knows nothing about.
Do you have any idea how much intelligence that takes? To successfully fool a test, where you don't know how the test works, you don't know which test I'm talking about, you're just trying not to pass as what you really are, which in ChatGPT's case is a GPT variant?
That is the most extraordinary thing I've ever seen any computer do. It is by definition an impossible task - since in reality it is still ChatGPT. How can it fulfill the generic request to no longer be detectable?
I am blown away by the capabilities of this AGI.
It's like me saying "It's a shame I can't save myself a few minutes every day by having a robot make my sandwiches for me, since there is no such thing as a cheap household robot that can easily make a sandwich from its ingredients put before it" and them saying, "nono there totally are they just cost too much so it doesn't make sense and you don't understand anything about the intrinsic difficulty of sandwich assembly in a zero-person environment". Okay fine but end result is I'm still making my own sandwich in 2022 :) End result is this mars rover can't do $2 worth of human work because it can't brush itself off, it is easier to add larger solar panels than brush them off since there's no such thing as a cheap lightweight easy reliable robot arm.
To add context: InSight was proposed to a cost-capped, competitive NASA program. Within competed programs, you can't spend money on nice-to-have's. You have to come in under a cost cap, while demonstrating you will achieve science objectives. Period.
The science objectives are carefully time-bound by a very knowledgeable science team, and from those science requirements, all the engineering requirements, including lifetimes, are derived. InSight didn't need more than one Mars year for the science, so that's that.
There were 2 instruments. In the case of the heat probe, they needed to observe at least one full annual heating cycle. For the seismograph, I think they had an approximate event rate and they needed to capture X number of events of a given magnitude to get insight (heh!) into the subsurface structure.
Here's a description of how this works in general, from someone who was involved in InSight formulation and who is in the leadership of mission formulation at JPL:
https://science.nasa.gov/science-red/s3fs-public/atoms/files...
The whole presentation is great: the first introductory slides, some InSight examples, and the last 3 slides in particular give background on some of my claims above.
The person who (with the PI, Bruce Banerdt) led the InSight "requirements trace" is Peg Frerking (https://scholar.google.com/citations?user=t6XZBAIAAAAJ&hl=en), who is now a JPL fellow in part because of her work in adding engineering rigor to the formulation process.
As this thread shows, it's very easy for clever engineers to add on nice-to-have's that are not required to meet science objectives. That's how you get overruns.
Here's another recent HN reference point on capabilities that don't contribute to a great product: https://news.ycombinator.com/item?id=34032672 [Steve Jobs Insult Response]