Topic: Repairing AI for Environmental Justice
Speaker: Prof. Dr. Aimee van Wynsberghe, Alexander von Humboldt Professor for Applied Ethics of Artificial Intelligence at the University of Bonn
Abstract: Let us imagine that Artificial Intelligence (AI) is broken. Not in the physical sense in which pieces are falling apart and need to be put together; rather, in the metaphorical sense in which there are serious ethical concerns related to the design and development of AI that demand repair. In this talk I will outline a definition of Sustainable AI as an umbrella term to cover two branches with different aims and methods: AI for sustainability vs the sustainability of AI. I will show that AI for sustainability holds great promise but is lacking in one crucial aspect; it fails to account for the environmental impact from the development of AI. Alternatively, the environmental impact of AI training (and tuning) sits at the core of the sustainability of AI, for example measuring carbon emissions and electricity consumption, water and land usage, and regulating the mining of precious minerals. All of these environmental consequences fall on the shoulders of the most marginalized and vulnerable demographics across the globe (e.g. the slave like working conditions in the mining of minerals, the coastal communities susceptible to unpredictable weather conditions). By placing environmental consequences in the center one is forced to recognize the environmental justice concerns underpinning all AI models. The question then becomes, how can the AI space be repaired to transform current structures and practices that systemically exacerbate environmental justice issues with the consequence of further marginalizing vulnerable groups.
Prof. Dr. Aimee van Wynsberghe is the Alexander von Humboldt Professor for Applied Ethics of Artificial Intelligence at the University of Bonn and director of the Institute for Science and Ethics and the Bonn Sustainable AI lab. She is co-director of the Foundation for Responsible Robotics, a member of the European Commission’s High-Level Expert Group on AI and member of the World Economic Forum’s Global Futures Council on Artificial Intelligence and Humanity. She is a founding editor for the international peer-reviewed journal AI & Ethics.
In each of her roles, Aimee works to uncover the ethical risks associated with emerging robotics and AI. Her current research brings attention to the sustainability of AI by studying the hidden environmental costs of developing and using AI.
Good afternoon so yes I am Professor Amy Van weinburg I’m the Alexander Von Hult Professor for the applied ethics of artificial intelligence I’m director of The Institute for Science and ethics and uh founder of the Bon sustainable AI lab so I’m over on the boner talg boner talg
Zees and um yes I I have so I’m I’m hoping that my talk today will resonate with some of you I already have some collaboration with se but uh I I think I’ve been here for almost three years and it’s amazing to me how massive the university is that there are so many
Different institutes and centers and departments and faculties and whatnot so I think this is a great moment to um to at least tell you a bit about my work and that myself and my team were always open to collaboration so if you’re interested then please follow up with me
Afterwards okay so today I’m going to give really a kind of a general overview on the research that I do as sort of an introduction to this is what’s happening at another Institute here at the University yeah so I’ll be talking about sustainable artificial intelligence and
This is within the field of AI ethics um but I’m what I’m trying to do with my research and my research lab is it to introduce this third wave of AI ethics which I’ll explain so a little bit about me just so you have a an introduction to
Me um I have a background in cell biology so I’m not a classically trained philosopher or or ethicist um and while I was studying I also come from Canada and while I was studying cell biology I worked at a robotics Institute where we were um training surgeons how to use surgical
Robots so this is sear Canadian Surgical Technologies and advanced Robotics and um it was this experience where I was kind of struck by the the way that medical Technologies were implemented and the way that they were developed that we were really looking at the efficiency of the robot rather than the
Bigger picture right rather than what was the nurses perspective how did the surgeon feel because they’re now I’m not sure if you’re familiar with surgical robots but the the surgeon is in this console yeah and they’re performing the surgery like this so they don’t touch the patient at all so I was curious
About you know what does this mean for the surgeon’s experience and it was this experience then that led me to look into ethics ethics of Technology how do we evaluate technology how do we design technology so then um I went on through my academic career and um also began
Working with Nos and with policy makers so I was a member of the European commission high level expert group on artificial intelligence and this was the precursor to the regulation that will be established here in Europe on artificial intelligence and um then coming here to
To Bon as the um as one of the Alexander Von Hult professors I shifted my research from the more traditional AI ethics issues to then broaden the perspective of AI ethics to look at environmental justice issues and so in each of these contexts whether it’s in Academia policymaking industry I’ve also
Worked in in Industry I take a particular um uh perspective or or Vision on what an ethesis is meant to be or meant to do and I refer to this as the ethesis as a designer so that the ethesis becomes a part of the design team of new technologies and the reason
Why the reason why I try to emphasize this is because for the most part ethics is usually considered an add-on at the end yeah you create and we see this a lot with artificial intelligence you create the algorithms you develop these models and then afterwards you think oops privacy right oops transparency
Fairness so the ethical issues come at the end and then then we have these stereotypes where the ethesis are the ones who come into the room and they’re like no you can’t do that you can’t do that so there’s this you know um tension where many developers think that ethics
Um stifles Innovation but what I’m trying to do is that ethics can actually be a part of the Innovation process if we start to to understand the ethesis as a part of the design team so um now getting into the the content of my talk I thought I would
Just briefly begin with what is ethics in general and yeah I could spend the next or the rest of the day talking about this but I won’t but just so that we’re all on the same page yeah so depending on who you ask you’ll have different answers to the question what
Is ethics some people focus on you know character traits of an individual virtue ethics how does one develop the virtues to become a good person um others will talk about theories of right actions right it’s the consequences of your action that determines whether or not it
Was good or bad or right or wrong or whether or not you follow certain principles and do duties but the uh vision of Ethics that I like to to use in my work is an Arista tilian vision and this is um the the idea that ethics
Is the study of the good life what it is and how we can achieve it and then we use the language of values to talk about the good life right and so then if we can achieve values like fairness privacy dignity sustainability then we come closer to being able to achieve the good
Life on an individual level and on a societal level okay so then with that understanding the ethics of artificial intelligence is in my in what I suggest it is is what is the impact of the design the development the usage and the scaling up of artificial intelligence on
Our ability to live a good life and so then looking at it in sort of like a linear way how does AI impact these values and then the impact on the values um resembles or has an impact on our ability to achieve the good life so super super General but I hope it’ll
Become um clear as I go through the rest of my talk also let me know if I’m speaking too quickly is am I speaking too fast because I’ve had way too much coffee today okay you know stays okay so then what I also want to do is to give
You some background information on this space of AI ethics yeah that it’s this massive field there are so many different ethical issues there are so many different values and ways that AI impacts these values sometimes promoting the values allowing for greater efficiency in a company for example allowing us to predict um consequences
Related to climate change resource distribution biodiversity so there could be positive impacts on the values and there could be negative impacts on the values and um I like to also discuss different Trends in the AI ethics field and I call them waves of AI ethics and
So the first wave that I suggest is the ethics of super intelligence of artificial general intelligence and within this wave then we had um you know Stephen Hawking almost 10 years ago was talking about how AI was going to end mankind right AI or we have the the Terminator scenario this is often
Invoked in the media that AI is going to rise up and just kill everyone yeah and so there’s these very like d day scenario we also have a a Revival of this kind of thing there’s been many statements now talking with with generative AI about how AI poses the
Biggest threat to extinction of humanity so there’s these um there’s a certain sort of uh category of ethical issues related to a technology that doesn’t exist so we don’t have super intelligence yet and we might not ever have it yeah but the ethical issues are related to this future technology
They’re very broad very general and they are dystopian AI is just going to wipe us all out this is what I would classify as the first wave of AI ethics then as a reaction to that once once uh Ephesus started to understand that this technology isn’t necessarily this Future
Super intelligence but we use it regularly I don’t have my phone but you know we all carry AI around with us on our phones unless you have one of the the flip phones which Yay good for you if you do have that um I’d love it if I
Could do that um anyway so we had ethicists who were then understanding that AI is very much a part of our day-to-day lives and that there are ethical issues related to very specific applications and is there a way that we can design it differently okay and so then what I mean
By this is we saw many examples so you know for example the self-driving Uber that killed a pedestrian and this raised the question of responsibility and how AI was impacting our concept of responsibility and again think about right good life values yeah we need security when it comes to who is
Responsible if somebody dies or who is liable if somebody dies during an in interaction with a technology yeah that’s how we feel we have a sense of security and safety about the country or the city that we live in and when this accident happened it was fascinating to
See how no one wanted to take responsibility and it was unclear who or what would be responsible the company said well the driver is responsible obviously and then you know the driver was thinking it’s a self-driving car right the whole point of it is there
There’s meant to not be a driver so how could the driver be responsible and S then they went to know the developers must be responsible right the computer engineer the computer programmer and they said it’s impossible for us to predict every single scenario that the car will engage in that’s why it’s
Machine learning yeah it learns as it as it develops so the developer can’t be responsible and the company certainly didn’t want to be liable so then we’re in this interesting situation who is responsible so this technology was challenging our foundations of the good life now to get into more uh specific
Examples and to also show there’s there’s so many things to talk about but I thought I would cluster them there are ethical issues related to Ai and data and many many many issues but if we think before we’ve even gotten to the stage where we are training the AI we
Have different issues related to moments in the life cycle of the data collection we learned through the Cambridge analytica Facebook Scandal about how data is sourced right the creation of a fake survey to generate information on individuals then this is used to Target them for political advertising so
Ethical issues about how you actually go about extracting or or getting data from someone that’s different from the collection that’s you know the creation the making somebody answer something uh and the answers are going to be used for something entirely different there’s a um the uh intuition of Deceit there yeah
So data sourcing then data collection are you familiar with clear view that company yeah so they they uh scrape the internet for um uh pictures with faces and then they use all of these pictures so Instagram Twitter Facebook I don’t I’m not like hip anymore so I don’t know what else is out
There Tik Tok yeah I’ve heard my kids say that yeah um so but uh scraping different social media sites to obtain images of faces and then using that to train an algorithm to be able to recognize faces right so aside from the problems with what kinds of images were
Were obtained it’s is it okay to to collect data in that way we refer to that as a secondary usage a a a lack of informed consent the people whose images were taken did not know that they were taken and did not give consent for this
Right so then we have you know that’s another set of ethical issues just related to how the data is collected we’re still not even at the the training data stage as we get closer to the training data stage then we think about how the data is labeled and so if we’re
If we’re creating a a machine learning algorithm for image recognition for example you need depending on the the methodology that you’re using you would need to be able to identify in a picture this is a cat versus this is an ice cream cone yeah and and you need people
Oftentimes to to label what the things are in an image in order for the training to occur but this can be very expensive right so now we see instances where they are using prisoners in prisons to do the labeling because it’s free labor they use refugees in refugee
Camps camps to do the labeling and they’ve created massive data labeling Farms they call them on other continents where people are working for you know a dollar 50 or um a Euro 50 or something like that a day and so it’s almost reminiscent of the um uh clothing cycle
Yeah you know what I’m saying um where we had you know children and these slave like labor conditions for people who were involved in the clothing industry so that we could have cheap clothes yeah so we’re seeing a kind of a repeat that’s happening in the AI space then we
Get to once the data has been labeled then we look at what happens when we’re training an algorithm not sure if you’re familiar with the saying garbage in garbage out okay so but this is the idea that um you know we often hear in the
Media that the AI is biased yeah the AI has biases but it’s that’s not the correct terminology we have biases within Society we have discrimination that happens within society and this is reflected in the data that we have and this is what is used to train the
Algorithm and so then these biases are exacerbated sometimes or are they are prevalent they are present in the output that we see coming from an AI so garbage that we have within Society that’s I don’t that’s not a very nice thing to say but you know the the biases and the
Discrimination that we have in society are unfortunately replicated and exacerbated when we’re using AI this is another problem related to the data another set of issues has to do with privacy and so um and this gets into you know when I’m talking about the values and the impact that the technology has
On the values um AI is changing our concept of of privacy when we bring AI into our home this was traditionally a private space and it has now become somewhat of a public space right because we have stories about how the AI is always on is always listening so what is
What is AI doing to our uh understanding of the concept of privacy or disclosure of new kinds of information now this can be positive or this can be negative AI can be used to make a prediction about an individual and this can be positive in the sense that um if somebody is
Coming from a country where they don’t have credit scores uh then they don’t have the possibility to be uh a full member of society if they’re moving to a country where a credit score is necessary yeah so they could use something like search history to create a kind of credit score
Um and and then allow them to become yeah a fuller member of society in this way but on the flip side this can also be used in a negative way that you have companies that are using this possibility of predicting about individuals um in less than ethical ways
I guess we would say that a company can scan all of the emails that are sent by an individual and this can be used to predict whether or not the individual is thinking about quitting for example and then what does the company do with this information do they want to fire the
Person before they quit or predicting suicide right there’s also um certain algorithms that are used to read the text read tweets read things like that to predict whether or not an individual is suffering from a mental disorder and might be on the verge of of committing
Suicide so so it can go what I mean to say here is that it can go both ways this idea of disclosing new kinds of information and excuse me how do you give consent for this right the developers don’t even know exactly what kinds of information they could uh
Reveal about an individual and so should we should we be in a position where we say yes you can use my my data for this when we don’t even know what could be the outcome of it so it then challenges our concept of consent as well as informed consent because oftentimes um
In the negative scenarios I presented nobody is asked whether or not their emails can be read Ai and transparency so it’s important to um Al how familiar are you with with AI I realize I’m talking about AI ethics but do do you work with with AI with artificial intelligence maybe
Some of you no so um important to to see it AI is this umbrella term artificial intelligence and it’s the idea of wanting to create a system that resembles human intelligence in a certain way maybe that’s making a prediction um maybe that’s uh understanding patterns in data in
Language for example and then there’s different ways of achieving this uh simulated this artificial intelligence you have good old-fashioned AI symbolic reasoning machine learning neural networks deep learning yeah and um Within These different method methodologies there are different concerns of course that are raised so with um with the older methods like good
Oldfashioned AI or symbolic reasoning it was clear what the rules were to um to develop the model so if uh you wanted to teach a model to be able to uh make a certain prediction you you gave limits to the model and and these were understood and known by the developers
With machine learning and this is now people mostly equate machine learning with with AI but with machine learning it introduced this new problem this new conundrum that the whole point is that the AI can learn something beyond what we know beyond what we understand so it can make predictions that we don’t even
Know are possible but that means in order for that to happen that we don’t set the boundaries we don’t give the rules we don’t say this is a cat this is a dog we want the machine learning algorithm to learn that uh itself but this means that the developers often
Don’t know or understand the rules and this is where the conversation comes in about the black box of the technology that you have inputs something happens and you have an output and one of the first examples where we learned about this was when um there was a company training an
Algorithm to uh predict or to recognize a wolf versus a dog and so they trained the model uh with the data that they had this is machine learning so they didn’t give it rules then they gave it a new set of data and they found that it was
Actually pretty good at predicting at at being able to identify whether or not the image was a wolf or a dog but then they did some reverse engineering right so they were trying to this is this is when we talk about explainable AI they were trying to get the system to explain
What pixels you know what part of the picture it used to make its uh prediction that it was a wolf or a dog and they found out that snow was the important factor so in all of the pictures where there was a wolf there was snow in the picture and in the
Pictures where there was a dog there was oftentimes not snow and so I mean it seems like a you know it’s like a that’s really interesting hadn’t thought about that but then when you get into more serious examples like that’s funny but then you get into a situation where you
Have you know the cartoon is meant to is also meant to be funny but if you think about it for for a second we could have a situation where we have these autonomous cars we don’t know how a decision was arrived at and um going back to what I talked about with the
Garbage in garbage out and the the prevalence or the the problem of of discrimination and biases in our society that’s the data that’s used to train the algorithm we don’t know what the rules are that are used to generate an output and then we have a situation where you
Know when it’s used in predictive policing for example already shown that people of lower socioeconomic status or people of a certain uh color or culture they are more often thought to maybe commit a crime in the future or their uh neighborhoods are policed more heavily because the algorithm is
Predicting that there could be a crime in this neighborhood or that that person could commit a crime so this becomes incredibly problematic this um this problem of where the data is coming from or the fact that that we’re replicating these biases and then not even knowing you know what is what is essentially
Happening okay so now getting to I know I’m throwing so much information at you but um getting to the sustainable AI part what I what I found um in my research in the so I was most often doing research in the second wave of AI ethics looking at specific applications
And raising these issues what what is the problem when you have the biases and the lack of of transparency in the algorithm you put these together and you have a problem but I found that uh in in recent years um I was struck by this
Idea that uh who I mean whoever did the marketing is a genius but you know when you hear Cloud ah the cloud it’s on the cloud yeah or AI you know what is it it feels intangible it feels like not of this planet that there is no physical
Infrastructure but there actually is a massive physical infrastructure that isn’t taken into account and so what I am suggesting with my research is that yes all of the ethical issues that I just laid out are very important but there’s another set of environmental justice issues that have not been
Addressed and um that it requires a more structuralist approach so rather than looking at n application a design problem looking at AI like this this massive massive thing of AI so I’m introducing sustainable artificial intelligence and this idea of of uh bringing sustainability into AI exploring what sustainability means in
The context of AI and with sustainability you often have the the connotation of um maintaining and what I want to show is that um this infrastructure of AI this whole this this massive thing that AI is is broken in a metaphorical sense and we need to
Repair we need to repair the the the environmental justices that I will um that I will raise so it’s not when I say sustainable AI it’s not just how do we keep this cycle going of exploitation it’s how do we understand this vicious cycle and decide what needs to be
Sustained or maintained and what needs to be repaired and rethought so I suggest that sustainable AI is a movement to Foster change in the entire life cycle of AI products towards greater ecological integrity and social justice and so I emphasize the entire life cycle of the AI product going back
To the idea um remember when I I was talking about how uh ethics is often this afterthought you know this thing that comes after and when we see when we look at the environmental justice issues that that I put on the table here or that I’m trying to uncover and raise
Awareness about we can see that we have to have this discussion and this mentality of sustainability before we even start collecting data right at the very very beginning how do we structure this infrastructure of AI so that we don’t maintain the exploitation of people in planet um then another part of this
Field of sustainable AI is um I’ve suggested that there’s these two branches there’s AI for sustainability and this is massive this is you know us using AI to achieve the sustain sustainable development goals and things like that and then there’s this sustainability of AI and it might seem
Like a no-brain or like oh yeah I mean of course but nobody was paying attention to the sustainability of AI and you have this massive movement I mean almost every big tech company go to Facebook Google Amazon they all have this like AI for social good AI for the
Planet Ai and it’s almost like a ethics washing it’s almost like a lip service you know we can focus on how we’re using AI to do these things and then you we don’t have to pay attention to to the other stuff and so what I’m trying to
Say is sometimes I even go um as far as to say if you don’t focus on the sustainability of issues there’s no there’s no way of achieving AI for sustainability these two things need to come together which companies do not like to hear by the way um okay so then um so
Just repeating what I had said before there is this massive growing effort to talk about using AI for good and then I’m trying to uh push back on this and say yeah I mean of course we want to use the technology for good but we have to understand the complexity of what what
That means and also using an unsustainable technology to achieve sust sustainable development goals seems counterintuitive no it seems like a bit of an oxymoron cognitive dissonance let’s use something that’s going to make the climate crisis worse to solve the climate crisis you know what I mean like
What um okay so then giving you some more examples of what I’m talking about and I I am watching my time so I I’m getting I’m getting there giving some more examples of what am I talking about when when I you know raise environmental justice issues I’ll just point out a
Couple here um in terms of ecological footprint and the first one is carbon emissions so when you do this training you um require energy to do the training and there’s often um carbon emissions as a result of this uh of using the energy and of course it depends on what type of
Energy you you’re using Renewables versus fossil fuels but traditionally it’s been fossil fuels and one of the first studies that was done um uh which kind of like struck a cord or spurred up this whole carbon emissions debate um showed that training one one natural language processing model on one
Computer one GPU resulted in 600,000 pounds of carbon dioxide which was the equivalent of of the carbon emissions of five cars over the lifetime of their of the car yeah so the car driving every day for for five years the training of one using one GPU one algorithm and this
Is you know if you have a scale of uh of um how many gpus are necessary how much how long they’re running how often they get retrained because you can’t just train an algorithm once you have to retrain it if you have a new data set or
If you have an have an update this 600,000 pounds of carbon dioxide isn’t even on the same scale if we’re talking about g chat gpt3 for example so I’ve been in many interviews where the Press will say oh Amy it’s just five cars like it’s not that’s not the end of the world
But what you know when I started to talk about the sustainable the the infrastructure of AI it’s not just one algorithm here and there right that we’re talking about an incredible number of algorithms an incredible number of gpus data centers that are um a part of this technology so then we’re talking about
Millions of cars being on the road and at a time when we need to reduce carbon emissions it seems it it seems counterintuitive that we’re we’re increasing carbon emissions through the the the training and the usage of AI um another ecological footprint to to uh to mention is the mining of Precious
Minerals and so again going back to this idea that there is a physical infrastructure to AI right it’s not it’s not it’s not in the clouds it’s on the planet and you need in order for this physical infrastructure to to function you need Precious Minerals you need gold
You need lithium you need so so many minerals and often these minerals are coming from mines outside of our continent and the mining conditions are awful and um now they started to do you remember Blood Diamond remember yeah so now they refer to the mine the the mineral minerals as blood minerals
Because we’re we’re seeing ourselves um again repeating history and um exploiting people in order for us to have our cheap AI systems on our phone or for us to you know have chat gpt3 and convert our CV into a poem and that kind of thing so the mining conditions is one
Thing how we are exploiting people but then you also think about the surrounding communities so I have I have researchers from Ghana on my team and um one of them is specifically doing research into the future of mining in Ghana because gold uh most gold is
Coming from Ghana and he talks about the mining communities and what the impact of mining does to the community in terms of um changing the soil and toxicity levels of the water and then I have another researcher from Ghana and she’s looking at um she’s looking at how women
Are particularly impacted by this so for the mining Community if your water is um less less than ideal the women are the one who have to walk further to go get clean water right uh or if they cannot if they cannot create a small plot of Agriculture they’re the ones who
Who have to figure it out so I I say this also to highlight or to point out that when we when we have this field of sustainable AI or when we have our focus on environmental justice issues it opens our eyes to Hidden Dem demographics that
We hadn’t been thinking about and so in the second wave of AI ethics we’re focused on you know privacy are you collecting my data that’s that’s almost a luxury ethical concern when you look at the other ethical concerns related to the uh environmental issues so um the the last ecological footprint that I’ll
Raise is electronic waste and so again reinforcing this idea that there is a physical infrastructure to artificial intelligence now we talk about um you need the latest gpus every the latest computers and servers and software and whatnot um every six months between two years that’s when you need to be um
Renewing them or replacing them sorry and um with that replacement you have to do something with the old ones right and I mean the I imagine I don’t have to explain to this group that electronic waste is often not kept on the continent where where it’s been used and again we
Have a situation where um we are uh increasing or introducing uh toxicity into communities outside of our own and so the toxins from the electronic waste go into the soil impacting the physical health of the communities around the mountains of electronic waste and then again impacting their ability to provide
For themselves in terms of clean drinking water or um when the soil is contaminated what what do you do about um growing food yeah so again reinforcing this idea that there are hidden demographics people whose voices are not heard who are being exploited for AI to be available in the US in
Europe in China in Russia and whatnot and then this this is you know I haven’t even got into the the use of water or the use of land and um water in particular in Taiwan the leading you know the leading producer I think it’s more than 50% of the chips are made
Again another necessary piece for the physical infrastructure of AI more than 50% of chips come from Taiwan and you need to have the cleanest possible water in order to um clean the chips and so now Taiwan is in a situation where they need to purchase water from other
Countries for the citizens to drink because they use the water for cleaning the microprocessor the chips yeah and so this so it goes yeah it goes uh on and on I could do more examples but I am watching my time so why care about the sustainability of AI I hope that I just
Answered that in the in the the previous slides but then just to sort of reiterate as I said before it’s not just uh Google Amazon Facebook uh you know the you know it’s not just these these select companies we see new companies popping up daily you know the billions
That governments are directing towards the development of AI on a federal level on a on a national level it’s it’s growing it’s not something that we’re going to be saying goodbye to uh not only that but another argument that I that I’ve been trying to make in my
Research is that um AI both relies on existing infrastructure so you need the water supply from a city and then of course you direct the water supply towards cooling the servers but you need public infrastructures that are already established in order to build in order
To develop AI but not only that now most companies use AI uh as their business model yeah they they use AI is their business so AI is becoming its own infrastructure yeah that it’s it’s not just relying on what we already have in society but you think of um think it was
A year and a half ago in the summer Facebook crashed for a day and there was so many companies that couldn’t function companies often in the global South that that couldn’t function because they needed the platform the AI platform in order for for their company for uh to
Work so yeah with that again I’m trying to invoke this idea that it’s not just one algorithm here and there it’s not just the problem of predictive policing or using Ai and recruitment of individuals for jobs it is this massive problem of just AI in general and that
Um the technology is growing and our investment in the technology is growing and then all of this to to also say uh and again this is not this isn’t the audience that I need to remind but that we are in a a global climate crisis and so you know talking about these two
Branches that I had and this in sustainable AI the of sustainability of and the and the sustain AI for sustainability and that you have all of these companies pushing towards you know AI for sustainability that at a time at the moment where we are at where this is
Code red for Humanity It seems impossible that we would move forward with this technology without bringing the two branches together and that what I think this means is not just a focus on carbon emissions not just um looking at the energy consumption of course this is relevant but when you really get deep
Into it and you understand where the minerals are coming from you understand what decisions are being made for the the drinking water on on different countries then we need to rethink this whole plan of AI before we push forward with the infrastructure in the way that
We have been doing okay so just to recap um uh the main thrust of my argument so I I started out you know talking about giving an overview of AI ethics and how there are so many different ethical considerations but then what I what I really wanted to emphasize and and I
Hope I did is that AI development and use causes environmental damage and that it’s time for us to now prioritize um and focus on what is this environmental what what are the environmental consequences and that that when we do that when that becomes the glasses that we put on to then look at
What are the ethical issues then our eyes are opened to continents to demographics that are not being heard or paid attention to when we talk about AI in Europe or AI in Germany so I’m saying our sustainable AI glasses are are necessary and should be this new era of AI
Ethics and that um so right now I think it’s very promising that we have regulation coming forward and even just this week we had many uh countries gathered in the UK uh talking about regulation so the US is now stepping up and they want to they they’re um approaching regulation as well Europe
Has been the front runner Canada Brazil China has even come out with AI regulation but what we still or what I still find is um uh there’s a lack of attention to are the environmental consequences that the really the focus is on is on priv privacy and fairness
And I’m not saying we should disregard that but it is time now for our governance mechanisms to include attention to and requirements for trying to repair this broken or vicious cycle of AI that I was talking about okay so I went a little bit over my time sorry for that and thanks for
Letting me sit down it was one of those days and um yeah that’s that’s it for me I’m happy to uh to take any questions thank you