AI Researchers Tout their Wares to the Climate Science Community

Climate modelers do it with digital crytals balls

Guest essay by Eric Worrall

Fantasy problem meet fantasy solution.

Here’s how AI can help fight climate change according to the field’s top thinkers

From monitoring deforestation to designing low-carbon materials
By James Vincent  Jun 25, 2019, 8:02am EDT

The AI renaissance of recent years has led many to ask how this technology can help with one of the greatest threats facing humanity: climate change. A new research paperauthored by some of the field’s best-known thinkers aims to answer this question, giving a number of examples of how machine learning could help prevent human destruction.

The suggested use-cases are varied, ranging from using AI and satellite imagery to better monitor deforestation, to developing new materials that can replace steel and cement (the production of which accounts for nine percent of global green house gas emissions). 

But despite this variety, the paper (which we spotted via MIT Technology Review) returns time and time again to a few broad areas of deployment. Prominent among these are using machine vision to monitor the environment; using data analysis to find inefficiencies in emission-heavy industries; and using AI to model complex systems, like Earth’s own climate, so we can better prepare for future changes. 

The authors of the paper — which include DeepMind CEO Demis Hassabis, Turing award winner Yoshua Bengio, and Google Brain co-founder Andrew Ng — say that AI could be “invaluable” in mitigating and preventing the worse effects of climate change, but note that it is not a “silver bullet” and that political action is desperately needed, too. 

Technology alone is not enough,” write the paper’s authors, who were led by David Rolnick, a postdoctoral fellow at the University of Pennsylvania. “[T]echnologies that would reduce climate change have been available for years, but have largely not been adopted at scale by society. While we hope that ML will be useful in reducing the costs associated with climate action, humanity also must decide to act.

Read more: https://www.theverge.com/2019/6/25/18744034/ai-artificial-intelligence-ml-climate-change-fight-tackle

The new paper is available here.

While I’m concerned that the potential of AI is being somewhat over hyped, there is no doubt AI could be a significant labor saving device for grant seeking climate scientists.

For example, below is my personal contribution to climate AI – a random climate psychology paper generator (press the button to see a new generated climate psychology paper) which I offer for free to the climate community.

What do you think? Would it pass peer review?

Update (EW): Stephen Skinner highlights an excellent reason why black box AI solutions are potential sources of extreme embarrassment for researchers.

Advertisements

58 thoughts on “AI Researchers Tout their Wares to the Climate Science Community

  1. Well, “Feminist glaciology” did get published, so Eric’s parody might get past the gatekeepers as well.

      • There is quite enough “artificial intelligence” and “virtual reality” in the climate community already.

      • Using artificial intelligence to find artificial solutions to an artificial problem : makes perfect sense to me.

    • Lesbian sheep is also in the peer-reviewed scientific record. I KID you not (pun intended)! We’re talking actual animal husbandry here, about real, actual sheep. Not the metaphorical, shy, closeted or merely bi-curious human women who won’t make the first move, and who need a proverbial lesbian Border Collie directing her about in order to get things started. In the paper, that author noted that sheep signal their sexual receptiveness by standing still. Hence, all those sheep standing out in the pasture could very well be lesbians, signaling that they’re desirous of another female sheep who is also standing still just a short distance away. The author did admit to a large uncertainty, and admitted that we can’t really tell. And this is a real science paper.

      What was the advisor or funding agency of the lesbian sheep paper thinking? Hmmm, lesbian sheep, sounds kind of sexy? I wish I were a lesbian sheep? I wonder if Title IX covers sheep? I could start to transition and make the cover of Rolling Stone?

      Which got me to thinking, what about sports teams named after male studs? Rams (male sheep), Bucks, Bulls. At the pro level, where there are only men involved, that’s just fine, but in the collegiate environment, we have Title IX. What are the women’s teams in such colleges called? Are they forced to be verbally raped by such an indignation as being referred to as male stud animals? How is this allowed to stand? The CSU Rams makes sense for men’s teams, but what about their women’s volleyball team? The Lady Rams? That’s just stupid. Ramettes? That sounds almost obscene. Well, maybe not for the girls softball team. Anyway, I leave it to activists to fix this pressing, crucial societal problem. And we”d better fix any cascading naming problems of cheerleader squads cheering for such teams, too, while we’re at it.

      Which got me to thinking: the human variety of lesbian sheep, so named because they behave somewhat like their pasture-ized sisters, standing still and receptively waiting, perhaps were the inspiration for Nuke Laloosh singing “women get woolly” on the bus in the movie “Bull Durham,” a movie about a strange love triangle taking place on a minor league baseball team named after an animal of decided male sexual prowess. Crash corrects Nuke about the lyrics, but maybe Nuke was right all along? Think about that.

      Which got me thinking, do you suppose there are actual bulls who walk around the pasture thinking, hey, I was born a cow in the wrong body? Do you suppose they would want to be a trans-gendered cow, if they could? Although I know, generally speaking, what happens to bulls who play no role in procreating. They have their Rocky Mountain oysters amputated, and then are fattened up for the slaughterhouse and eventually become delicious. So maybe trans-cow is not a good option for now.

      I have written several derision tomes about feminist glaciers, and will not repeat them here. I will only reiterate… wtf? Do feminists or anyone for that matter, look at Puget Sound and weep for the loss of the ice that cut that valley? NO, THEY DO NOT! Sometimes, when the lefty idiots of Portland or Seattle do some stupid thing, I wish another Lake Missoula was filling up. But then I think, no, that’s unkind, and besides, at that point they’d all understand the power of nature and it’s ambivalence to our tender and often misguided feelings. Okay, that got way to serious. I’ll try to wrap this up.

      If you wonder what I meant by all these references to ideas around confusion and choice in sexual, gender and animal and ice identities, and my further disturbing eliminationist rhetoric towards bovines, I assure you, so do I. I’m sure some lefty of the Lewandowsky Oreskes love child variety would be more than happy to diagnose me over the internet and tell me what I’m thinking. And of course I mean here the lovely Naomi Oreskes of Harvard, Merchants of Doubt fame, and not her highly testosteroned brother Michael, formerly of NPR, before his Bull-ishness caused ME-TOO to slam dunk his career in “public” service. Not that there would be anything wrong with a Lewandowsky Michael Oreskes hookup, mind you. I’m as open minded as the next guy. It’ just that the latter would never lead to a love child.

      Now, let’s stop with all this satire and double entendre to get to a serious point. What would it take to convince you that science, as practiced within the research university or the federal bureaucracy was going off the rails, being taken over by activists, turning honest questioning of nature into tendentious self-serving exercises of the form of rent seeking or CV padding, carried out like a prosecution in a criminal case, wherein the conclusion of the prosecution is per-determined, and a case made for that version of the truth, only. Would science papers about lesbian sheep and feminist glaciology narratives be a clue to you? If not, they should be. Suffice it to say, dogged feminist activists abound in the Academy (pun intended) and non-scientist activists rule over most of what is called climate science in 2019. Politically, we must stop funding this crap.

      • On the subject of gender appropriate team nicknames: I recall seeing a sweatshirt featuring a female canine caricature with an armful of pups jumping up to kick a guy in the chops while stuffing a basketball.

        Similarly, should we not properly speak of male chauvinist boars and female chauvinist sows?

  2. Love it… That’s a better use of AI. It’s just pattern following more than pattern recognition. :<)

  3. Pass peer review.
    Definitely!
    There is another kind of Artificial Intelligence than being discussed currently.
    And it has been around for a long time, and it has to do with central banking.
    The system takes a run-of-the-mill PhD in interventionist economics. Who has learned by rote all about currency depreciation and interest rate manipulation and elevates them as Fed Chair.
    Skilled at FedSpeak he is instantly a genius because of the office and adored by the financial media.

  4. I wonder how IA will help us to solve a non existent problem … perhaps ingurgitating trillions of Mann made data ?

  5. Wait, what? You mean your personal contribution hasn’t been accepted yet? Looks as good or better (with all the fancy words) than most other accepted/peer-reviewed “studies”. 😉

  6. well theyre targeting the most gullible, so it prob would pass pal review,which then becomes peer reviewed
    and if you pay more you get more media time

  7. Anything that depends on a human to create can be made to be biased towards whatever the human wants. We should know this by the complicated and proprietary programs being used to screen speech. Artificial Intelligence is whatever the human says it will be. Nuff said.

    • Yes indeed Pamela:
      AI programs and algorithms contain the bias of the those that produce them.
      The main problem here is that it nigh on impossible to argue with an algorithm; particularly when hidden deep in the legal tangle of commercial privacy rights.
      An AI algorithm designed to produce propaganda is anathema.

  8. “The authors of the paper — which include DeepMind CEO Demis Hassabis, Turing award winner Yoshua Bengio, and Google Brain co-founder Andrew Ng — say that AI could be “invaluable” in mitigating and preventing the worse effects of climate change, but note that it is not a “silver bullet” and that political action is desperately needed, too.”

    More Star Trek fantasy, without the 300 years of additional technology.

    How large is Google’s version of AI department? Facebook’s? Twitter? And many other tech giants who use AI to direct their users and to censor voices they dislike.

    Most, if not all, AI programs focus on simple binary choices to develop their AI (Alleged Intelligence). Even then, they rely upon many large teams of people to carefully watch and constantly update decision factors.

    Recent news highlight the nefarious actions tech giants has utilized in forcing their preferred often false choices.

    America, should avoid such a manipulative nightmare!
    Expecting AI to replace humans at “mitigating and preventing” is a pathetic attempt to substitute hard coded diabolical absolute bias for human’s changeable weak bias.

    • “Most, if not all, AI programs focus on simple binary choices to develop their AI (Alleged Intelligence).”

      That’s one of the most fallacious descriptions of AI I ever read.

      I do not even see what “simple binary choices” even mean in this context.

      “Even then, they rely upon many large teams of people to carefully watch and constantly update decision factors.”

      Currently, training neural networks requires some hand-tuning. Yes. But compared to the results it achieves, it’s clearly a very minimal amount of information that is of human origin.

    • AI as in neural networks or optimizing algorithms and heuristic algorithms. Not smart, some intelligence, and unimaginative. Enter the climate scientists and modelers to supply the missing links.

  9. …stand in need to unprecedented vexatious…

    Eh, whatsa matter for you, HAL? You not familiar with Charles Fillmore / case grammar? You smart, but you not go far iffa you not speak the English so good.

    • Correct grammar is not a requirement in psychology papers, especially climate psychology papers 😉

  10. “What do you think? Would it pass peer review?”

    A laughable joking question, obviously!

    After all, how can anyone ask such a question after over a decade of horrid pal reviews coupled with corrupt publishers subverting science.

    The real question is how much it will cost you to publish, since publisher pal reviews can guarantee choruses “yea” reviews.
    Only those papers that conflict with the “Chosen” beliefs have difficulty.

  11. Stalin is reported to have said that it doesn’t matter who votes, but who gets to count the votes. The Progressive Era analog is it doesn’t matter what the data are, but who gets to train the AI algorithm.

  12. Guess all the journals and other publications are going to have to deploy AI screeners to identify hoax papers. A difficult job, especially in psychology when over half the papers are unintelligible, misusing statistics, and ignoring basic scientific principles.

  13. What if – now bear with me – the AI shows there is NO problem? Crazy I know. But what if it cannot detect any “finger print” for anthropological warming? What if it finds temperature adjustments are wrong or that there is more UHI effect in the datasets than is allowed for? That would certainly be interesting.

    • Would be interesting to throw all known climate variables into a neural net and see if it decides that CO2 is the control know, or something else.

  14. AI integration would explain the rapid spread of “what if” and “question mark” research papers published in recent years in climate science warming extrapolation studies with standard inserts of agenda science and policy scare statements. The opportunities for push button academic promotion via volume-based publication efforts are endless. The algorithm may even have a name called the MMann button.

  15. Here’s how AI can save the planet

    Invent robots that convert CO2 to O2 using a synthetic photosynthesis to generate their power. (Plant-bots?)

    The robots would have AI powerful enough to do all your work for you. They would have to work at night because they would have to lay around on the lawn all day recharging. They could have retractable “leaves” for daytime charging. They could be programmed to move around every half hour so they don’t kill your grass. They could have sensitive eyes to work in low light conditions so you can cut lighting power at workplaces by 90%. They can work in cold or hot so no need for workplace heating or cooling. Since no commuting is necessary we can take down all the street lighting and billboard lighting, lighted signage, exterior building lighting, parking lighting, etc.

    Make 35 trillion of them and give five as slaves for every human. No human would have to travel to work as the robots can earn the money for you. So no more cars, passenger trains, or airplanes are needed for commuting.
    The extra robots not utilized can paint all the streets, roads, parking lots, and sidewalks white to increase albedo.

    Hey this is just as plausible as thinking we can run an electric grid on 100% unreliable solar and wind power.

  16. Left hand not talking to right hand…

    AAAS: Machine learning ‘causing science crisis’
    https://www.bbc.co.uk/news/science-environment-47267081

    “Machine-learning techniques used by thousands of scientists to analyse data are producing results that are misleading and often completely wrong.
    Dr Genevera Allen from Rice University in Houston said that the increased use of such systems was contributing to a “crisis in science”.
    She warned scientists that if they didn’t improve their techniques they would be wasting both time and money. Her research was presented at the American Association for the Advancement of Science in Washington.
    A growing amount of scientific research involves using machine learning software to analyse data that has already been collected. This happens across many subject areas ranging from biomedical research to astronomy. The data sets are very large and expensive.

    ‘Reproducibility crisis’
    But, according to Dr Allen, the answers they come up with are likely to be inaccurate or wrong because the software is identifying patterns that exist only in that data set and not the real world.
    “Often these studies are not found out to be inaccurate until there’s another real big dataset that someone applies these techniques to and says ‘oh my goodness, the results of these two studies don’t overlap‘,” she said.
    “There is general recognition of a reproducibility crisis in science right now. I would venture to argue that a huge part of that does come from the use of machine learning techniques in science.”
    The “reproducibility crisis” in science refers to the alarming number of research results that are not repeated when another group of scientists tries the same experiment. It can mean that the initial results were wrong. One analysis suggested that up to 85% of all biomedical research carried out in the world is wasted effort…”

    Didn’t see that coming?

  17. When the AI machine starts spewing out Nobel awards to Al Gore and GSA awards to Mann for legal maneuvers you will have a clue. Will also be an awards category for other algorithms like Google in the fraternity of AI big brothers.

  18. Great idea. AI could also be used to help fight space aliens and also, the coming zombie invasion (you know it’s just a matter of time). And if flying pigs ever became a problem – presto, AI once again to the rescue!
    The sky’s the limit.

  19. 1) We haven’t made any AI yet. We’ve got some clever scripts but “intelligence” means something, and it arguably means the capacity to reason generally, and the best we’ve got is a long way from that (obviously there are varying dictionary definitions).

    2) One of the most sophisticated neural networks around is supposed to be YouTube’s recommendation system, and it’s awful. In my experience it’s wrong 99.5% of the time, often ridiculously wrong. A trained border collie would do an infinitely better job. I mean, they might also be over-hyping it and it’s just a 900-line batch file, who knows.

    3) If the predictions don’t pan out, the hypothesis is still wrong, no matter how many lines of code you have.

    • Not an intelligence with a semblance of the degrees of freedom exhibited by even mediocre human beings. We have AI of the form of optimizing algorithms (e.g. neural networks), heuristic algorithms, and missing links filled with brown matter or “fudge factors” as we have come to expect in climate, political, and social sciences.

      • Slightly clever scripts. No intelligence in sight.

        I mean, people have kind of standardized on “intelligence” meaning “literally anything that continues without a button press,” so it’s hard to blame people for using it, even though it’s been diluted to the point where it means almost nothing.

        I just have a problem with a word meaning “almost any value on an entire spectrum.”

  20. I question the use of the term “AI” when it comes to defining these analysis. True AI would come back saying the whole global warming is a hoax and humans are wasting their money. Their version of AI is nothing more than modeling. I contend no machine is capable of true AI. There are aspects of human reasoning that no machine can duplicate. It is the God factor- made in his image that can not be matched.

    • AI is very sensitive to inputs. If you just loaded it with the last 150 years of data, it would see CO2 levels rising and temperature rising and likely conclude that CO2 was driving temperature rise (or that temperature rise was driving CO2 rise).

      If you told the AI that it was just as warm a thousand years ago the AI might produce more interesting conclusions. But plenty of human climate researchers manage to ignore the medieval warm period, so why would they give this information to their AI?

  21. For the guys that have no clue yet what A.I beauty stands for.

    In a mater and time of only one hour, an A.I does achieve to redo and replicate one of the most harder
    Human experiment in physics, a Nobel price experiments, with no clue what so ever given of any connections either in subject of physics or otherwise as an initial condition.
    A complete blind fold test of a replication… impossible in human terms to do.

    Not only that, but the handling of the particulars in the term of that experiment were in a much much higher tuning,
    and accuracy implementation by the A.I handling, than the humans that first produced such experiment in the first place.
    Where no problem for the A.I to keep doing it in much better and more efficiently,
    to a given proposition of the protocol to be considered as;
    a utilizing of it successfully as a tool, not only as a replication, but a lot more, a utilization as a tool,
    far beyond the human capacity and human skill to achieve, or even dream about, at that point.

    All this was proved to be achieved in concept, principle and other wise, against any odds, in only one hour,
    By an A.I…. not ever possible for humans, ever.

    In one only hour, an A.I not only achieves a proper replication in a very complex and in a complete blind fold condition,
    but shows clearly that it could even utilize it as a tool, beyond the experimentation point, in one of most difficult scientific experiments, a Nobel prize one in physics.
    (This is not a joke, it is a very real thing, mind blowing yes, but still greatly underestimated.)

    Please do stop underestimating the A.I,
    it is going to be the path of our future, for the best or the worse, which ever way you want to look it.

    A.Is, fully matured, do not make mistakes, we humanos do, a lot of shit mistakes.
    Open your eyes, A.I is a real and a very potential “thing”, which we can not afford to underestimate, as we could miss a lot in meaning of time and gains in our life.

    oh, ok…

    cheers

  22. quote “The authors of the paper — which include DeepMind CEO Demis Hassabis, Turing award winner Yoshua Bengio, and Google Brain co-founder Andrew Ng — say that AI could be “invaluable” in mitigating and preventing the worse effects of climate change…”

    He’s just putting his hand up and saying “I want some money too, and I’ll even pretend to find more problems”.

  23. Here’s how AI can help fight climate change according to the field’s top thinkers

    Turn off the computers.

    You’re welcome, please send my cheque to my Swiss bank account

  24. Climate modelers do it with digital crytals balls
    Guest essay by Eric Worrall –>

    Climate modelers do it with digital – crystal – balls
    Guest essay by Eric Worrall

  25. “Technology alone is not enough,” write the paper’s authors, who were led by David Rolnick, a postdoctoral fellow at the University of Pennsylvania.

    “[T]echnologies that would reduce climate change have been available for years, but have largely not been adopted at scale by society. While we hope that ML will be useful in reducing the costs associated with climate action, humanity also must decide to act.”

    Didn’t anybody pass the good news to the University of Pennsylvania:

    Nowadays there’s highly funded “super computers” for easy modelling climate scenarios – to restart till the result fits the mémé.

Comments are closed.