Machine Learning Helps Us Preserve The Environment! 🌲


Dear Fellow Scholars, this is Two Minute Papers
with Károly Zsolnai-Fehér. In this series, we have plenty of videos on
DeepFakes, and whenever I publish one of these, I hear a lot of doom and gloom, some of which,
is, of course, justified, but in this video, the pendulum shall swing on the other side. Today, I would like to show you that machine
learning has a tremendous capability to do good for us. Today’s paper is a large collaboration between
The University of Pennsylvania, DeepMind, Google AI, Microsoft Research and more research
institutions, and it discusses many high-impact climate issues and how we can use machine
learning to better preserve our environment. It can help with some things that I thought
of, and so many things that caught me by surprise. Note that this paper will not have the visual
fireworks that you may expect. In many of our episodes, you get ice cream
for your eyes, but today, you get an ice cream for your mind. Let’s dive in. The transportation sector is responsible for
a big chunk of carbon-dioxide emission across the planet, unfortunately, up to one quarter
of it, so, this must be our first topic. On the surface, improving on this sounds like
a fruitless endeavor because more than two thirds of the transportation takes place in
the form of road travel that relies on the high energy density of fuel. This leads to a great deal of pollution, but
on the other hand, this kind of transportation is also a necessity to make sure we have our
supermarkets and pharmacies well stocked. So how can we address this with machine learning? Well, transportation is, of course, necessary,
but it turns out, we can cut down the emissions substantially by using it to optimize vehicle
routing, increase the loading and more. It turns out, learning-based techniques are
excellent at estimating and predicting road traffic patterns, and are often even better
than handcrafted statistical methods at that. It can also reduce congestion on airports
and improve the efficiency of ride-sharing services. The total number of vehicles dispatched may
be reduced this way, and on the other end, the ones that have to be dispatched can also
be made more efficient. The paper discusses and points to a collection
of other works in which machine learning is used to design more efficient internal combustion
engines and better aerodynamic design for these vehicles. Of course, hybrid, electric and autonomous
vehicles are also on the rise, and machine learning is also a great deal of help in this
area – autonomously driven vehicles can communicate with each other and reduce traffic congestion. One of my favorite parts of the paper was
about the freight sector. It says that these autonomous vehicles can
perform a maneuver called platooning – this means that many trucks drive very closely
together to reduce air resistance for all the trucks behind the first one. You can see similar maneuvers in all kinds
of car racing competitions, where the pilot in the second position hides behind the first
to let it eat up the air resistance, and overtake it at an opportune moment. A proper implementation of this platooning
for trucks is only possible when there is continuous and precise communication between
these vehicles to make sure they can accelerate and break in an orchestrated manner. As you see, machine learning can be used to
make our lives and our future immensely better. But, is that all? No, no. Not even close. If you have a look here, you see that here,
we have only talked about one domain out of thirteen that the paper discusses. It is a great resource, very easy to read,
so make sure to have a look. This is not only for researchers and engineers. If you are a politician, journalist, an entrepreneur
or just a curious mind who would like to know more, make sure to have a look. If you do, you’ll not only see a well-crafted
work, but you’ll also learn how to make a statement and how to prove the validity
of this statement. This is a skill that is necessary to find
truth in your life. We need a little more of that these days. So, please, read your papers. If you think this is important, make sure
to share the paper or this video with your friends, we all have to know about the fact
that there is an immense amount of research work invested in crafting a better future
for all of us. However, there are, of course, other ways
to preserve the environment, and you can also be a part of that too! Please consider donating to Team Trees, a
huge YouTube collaboration that aims to plant 20 million trees all around the globe by January
1st, 2020. I think this is an amazing project and the
bottom line is that I and you Fellow Scholars should participate together! Make sure to head over to teamtrees.com and
chip in if you can spare a few dollars. Note that every dollar plants a new tree and
these are not just being put everywhere willy-nilly, but all this is done by the Arbor Day Foundation
and they make sure that they are planted where there is indeed a great need for them. Thanks for watching and for your generous
support, and I’ll see you next time!

62 thoughts on “Machine Learning Helps Us Preserve The Environment! 🌲”

  1. I am going to choose machine learning in my third year of CS thanks in part to the wonderful content you put out 🙂

  2. I would love to be apart of this project in some way. This is going to change the world forever and I want to be involved with that somehow.

  3. Would be interesting to know, how much energy (fuel) truck chains could safe due to this method. I mean every single bit of reducing CO2 is important. Trucks aren't that aerodynamic, so there should be the possibility to safe a lot, due to the fact there are millions of trucks around the world. I'm studying psychology in germany and i apreachiate, that you're presenting those kind of papers. Love it.

  4. "[…] the human race might easily permit itself to drift into a position of such dependence on the machines that it would have no practical choice but to accept all of the machines' decisions.[…] Eventually a stage may be reached at which the decisions necessary to keep the system running will be so complex that human beings will be incapable of making them intelligently. At that stage the machines will be in effective control. People won't be able to just turn the machine off, because they will be so dependent on them that turning them off would amount to suicide."
    – Uncle Ted

  5. Detecting deepfakes with machine learning which were created by machine learning (but machine learning is good?)
    This is like using Nukes to eliminate others who use nukes.
    Besides, It would reach a point where the deep fakes are so good that no even machine learning would be able to spot them, or it would take much more energy and time to spot deep fakes than create new ones, beyond that, there is not monetary incentive to spot fake news or deep fakes, but there is a lot of economic incentive to make those.
    We are the ones who design, take decisions and do stuff. Replace those steps and you are not solving things, you are just eliminating our purpose to exist.
    Now you give the control of the whole traffic to an IA, what better way to kill us all :).

  6. But also, do not forget the number of jobs that would be displaced by self-driving trucks, especially in economies such as the U.S.
    We must constrain our optimism about the future with discipline to ask ourselves "should we do it?" and not just "can we do it?"

  7. Platooning trucks is just a stupid version of freight trains. Trains are much, much more efficient, can be used with low-carbon electricity, can move much larger masses, with higher density. You only need truck for the last miles/kilometers.

  8. Ok, Machine Learning is amazing for many reasons, yes, but let's face it, the sheer computing power needed to train all those AIs is so demanding, I've heard something like 10% of current computing power of the entire tech industry is dedicated to AI machine learning. That leads to enormous amounts of power required and therefore, destroying the environment.

    Also, we wouldn't need so many trucks moving products if people were educated better and not using things like Amazon Prime 24/7. The real problem is a lot higher up the chain…

  9. The most important problem that should be considered as a high priority to get solved by the AI and also should be considered as a great danger for humanity on the earth, is overpopulation. Carbon dioxide emission of 500 million people (maximum standard population for earth) is much less and easily controllable than the emission coming out of the daily needs and activities of 8 billion people.

  10. Damn man I cant tell you how much I like your videos, I hope that there was more channels like this one. Good job.

  11. using ecosia as a search engine also HELP it plants 1 tree every 45 searches
    already 70 millions have been planted !!

  12. Stop polluting The Pacific Ocean and producing unnecessary billions of plastic bags, and THIS will help the environment much more!

  13. I have an interview at Cambridge soon and your videos have given me so much to talk about! Thanks so much for spreading all this knowledge!

  14. User: Reach destination safely while optimizing carbon emission.
    AI: Subtly induces traffic accidents in other road users, especially accidental deaths of Big Oil executives.

  15. I usually love your videos, but this one sends the wrong message. Designing better combustion engines and producing more cars (autonomous cars will eventually result in more car on the road) is a disaster for the environnent! We need to give up petrol and metal extraction of we want to have a chance!

  16. The Single biggest Co2 is over production of global Food left uncollected, waisted, and thrown away to Rot openly
    Hand's DOWN

  17. Just an opinion but if you put something related to the work even though work doesnt have any image or video to show, it would be better. I got distracted a lot but it may be just me

  18. Next step:produce a general purpose so to solve climate change and forget to put the condition of not exterminating humans

  19. Im sorry but nothing ist saved … Business dont Work Like that … If i can save 10% fuel i can use 10% more Trucks and make 10% more Money. Thats how it works and how they learned it. And the plant 20 Mil trees Project is nice but … Do the math and you See that 20 Mil trees dont do much … Thats the co2 ouput of only usa of a single day If i remember right. And the trees need at least 10 years until they big and of use. And i they die in lets say 100 years they Release the bound co2 again. As Long as Humans are selfish and greedy creatures there is No Hope for earth. Thechnology is the Problem Not the solution. It is Always Used for more, faster, cheaper, more effizient, …. Thats No Help …

  20. Don't worry about CO2 global warming as peak oil as expected in 2035. Though saving energy and materials is a good idea. Recycling as well. Otherwise the industrial era may collapse.

  21. As said, they will create an artificial superbrain and make it their God to bring peace and optimize the performance of the economy.

  22. All we have to do to save the environment is to do nothing. Stop messing with it. We don't need to plant trees or make some special bacteria to eat plastic. We just need to stop doing things and in 10 years all our problems will be solved.

  23. As it stands, efficiency dictates a company's ability to provide profitable services so removing those limitations will just open the doors to a greater indirect exploitation of the environment for profit. Basically, I feel like this will just lead to more transport rather than reduced emissions because of induced demand (the same reason why building new roads doesn't ease congestion).

Leave a Reply

Your email address will not be published. Required fields are marked *