AI's Hidden Cost: Is Your Chatbot Harming the Planet?

>> Well, to start off with, can you tell us >> Well, to start off with, can you tell us a bit about JCTM and and what the a bit about JCTM and and what the a bit about JCTM and and what the mission is? mission is? mission is? >> So, the Jewish Climate Trust is a new >> So, the Jewish Climate Trust is a new >> So, the Jewish Climate Trust is a new philanthropic organization. The mission philanthropic organization. The mission philanthropic organization. The mission is to maximize the Jewish people's is to maximize the Jewish people's is to maximize the Jewish people's positive climate impact. So that means positive climate impact. So that means positive climate impact. So that means the state of Israel, the diaspora the state of Israel, the diaspora the state of Israel, the diaspora communities, especially in North communities, especially in North communities, especially in North America. How can the philanthropic world America. How can the philanthropic world America. How can the philanthropic world best maximize the difference that both best maximize the difference that both best maximize the difference that both of those groups are making on climate? of those groups are making on climate? of those groups are making on climate? Both mitigation, reducing emissions, and Both mitigation, reducing emissions, and Both mitigation, reducing emissions, and adaptation and resiliency, dealing with adaptation and resiliency, dealing with adaptation and resiliency, dealing with the climate crisis we're already seeing the climate crisis we're already seeing the climate crisis we're already seeing everywhere around the world. everywhere around the world. everywhere around the world. >> Amazing. So uh today obviously we're

>> Amazing. So uh today obviously we're >> Amazing. So uh today obviously we're talk we're here to talk about the AI talk we're here to talk about the AI talk we're here to talk about the AI revolution and as a climate uh kind of revolution and as a climate uh kind of revolution and as a climate uh kind of geared organization I suppose I'm one of geared organization I suppose I'm one of geared organization I suppose I'm one of the things I'm interested in is uh what the things I'm interested in is uh what the things I'm interested in is uh what you're wrestling with when it comes to you're wrestling with when it comes to you're wrestling with when it comes to this AI revolution uh and its impact on this AI revolution uh and its impact on this AI revolution uh and its impact on the environment. the environment. the environment. >> For sure I think AI more than almost any >> For sure I think AI more than almost any >> For sure I think AI more than almost any other technology huge challenges huge other technology huge challenges huge other technology huge challenges huge opportunities and a lot of unknowns. Um opportunities and a lot of unknowns. Um opportunities and a lot of unknowns. Um so firstly on the opportunity side we so firstly on the opportunity side we so firstly on the opportunity side we know that there's scope for AI to help know that there's scope for AI to help know that there's scope for AI to help improve energy efficiency um to change improve energy efficiency um to change improve energy efficiency um to change the way that grids work to to discover the way that grids work to to discover the way that grids work to to discover new materials that could be used in in new materials that could be used in in new materials that could be used in in agri voltaic or photovoltaic agri voltaic or photovoltaic agri voltaic or photovoltaic materials um to for solar panels and all materials um to for solar panels and all materials um to for solar panels and all sorts of other innovations. But there is

sorts of other innovations. But there is sorts of other innovations. But there is a huge concern over the amount of energy a huge concern over the amount of energy a huge concern over the amount of energy that AI is using now and the amount that that AI is using now and the amount that that AI is using now and the amount that it might use in the future. So there it might use in the future. So there it might use in the future. So there have been a few reports actually have been a few reports actually have been a few reports actually published this year uh the MIT published this year uh the MIT published this year uh the MIT technology review from a few months ago technology review from a few months ago technology review from a few months ago and the international energy agency also and the international energy agency also and the international energy agency also >> who basically say we know that these >> who basically say we know that these >> who basically say we know that these models can use a huge amount of energy. models can use a huge amount of energy. models can use a huge amount of energy. >> We don't know that much more than that >> We don't know that much more than that >> We don't know that much more than that >> because the big tech companies that are >> because the big tech companies that are >> because the big tech companies that are building these models are keeping a lot building these models are keeping a lot building these models are keeping a lot of that information closed and of that information closed and of that information closed and proprietary. So they've run experiments proprietary. So they've run experiments proprietary. So they've run experiments on open source models to try and on open source models to try and on open source models to try and approximate the energy usage that a approximate the energy usage that a approximate the energy usage that a query to chat GPT or generating an image query to chat GPT or generating an image query to chat GPT or generating an image through midjourney or something might through midjourney or something might through midjourney or something might have. And they've come up with estimates have. And they've come up with estimates have. And they've come up with estimates but the big variables the number of but the big variables the number of but the big variables the number of parameters in the model and where the parameters in the model and where the parameters in the model and where the data centers actually are and where data centers actually are and where data centers actually are and where they're drawing energy from are things they're drawing energy from are things they're drawing energy from are things that we still don't know. So we know

that we still don't know. So we know that we still don't know. So we know that we know enough to be concerned. M that we know enough to be concerned. M that we know enough to be concerned. M >> we know enough to know that there's >> we know enough to know that there's >> we know enough to know that there's going to be a policy challenge, an going to be a policy challenge, an going to be a policy challenge, an infrastructure challenge for AI data infrastructure challenge for AI data infrastructure challenge for AI data usage and energy usage, but we don't yet usage and energy usage, but we don't yet usage and energy usage, but we don't yet know enough to be able to project that know enough to be able to project that know enough to be able to project that accurately. And so I think accurately. And so I think accurately. And so I think >> that comes downstream for an >> that comes downstream for an >> that comes downstream for an organization like us where we're focused organization like us where we're focused organization like us where we're focused on climate. We want to know how much of on climate. We want to know how much of on climate. We want to know how much of a climate impact our AI usage might a climate impact our AI usage might a climate impact our AI usage might have. have. have. >> But we don't, right? We don't see when >> But we don't, right? We don't see when >> But we don't, right? We don't see when we're sending queries to chat GPT or to we're sending queries to chat GPT or to we're sending queries to chat GPT or to Gemini or to any others what the carbon Gemini or to any others what the carbon Gemini or to any others what the carbon footprint might be and we have to infer footprint might be and we have to infer footprint might be and we have to infer that. So we know for example that a that. So we know for example that a that. So we know for example that a video will take a huge amount more of video will take a huge amount more of video will take a huge amount more of energy and processing power than a text energy and processing power than a text energy and processing power than a text response or an image response but we response or an image response but we response or an image response but we don't know exactly how that's going to don't know exactly how that's going to don't know exactly how that's going to develop. And so I think what we're develop. And so I think what we're develop. And so I think what we're wrestling with as an organization is a wrestling with as an organization is a wrestling with as an organization is a microcosm of what's going on in the microcosm of what's going on in the microcosm of what's going on in the general scene now, which is that general scene now, which is that general scene now, which is that everyone's trying to understand the everyone's trying to understand the everyone's trying to understand the implications of this for energy and implications of this for energy and implications of this for energy and obviously for climate and we don't have obviously for climate and we don't have obviously for climate and we don't have enough information to really to be able enough information to really to be able enough information to really to be able to project.

to project. to project. >> As a philanthropic fund, you work with a >> As a philanthropic fund, you work with a >> As a philanthropic fund, you work with a lot of NOS's. Uh as a generalization, lot of NOS's. Uh as a generalization, lot of NOS's. Uh as a generalization, how are how are NOS's looking at uh AI how are how are NOS's looking at uh AI how are how are NOS's looking at uh AI and working with AI? I think that and working with AI? I think that and working with AI? I think that there's a there's sometimes a lag in the there's a there's sometimes a lag in the there's a there's sometimes a lag in the NGO world um to adopt new technologies NGO world um to adopt new technologies NGO world um to adopt new technologies because you you need to sort of be because you you need to sort of be because you you need to sort of be frontooted and and innovative and and frontooted and and innovative and and frontooted and and innovative and and often they're very focused on the often they're very focused on the often they're very focused on the program side um and don't always have program side um and don't always have program side um and don't always have the bandwidth to understand how best to the bandwidth to understand how best to the bandwidth to understand how best to use these new emerging technologies. One use these new emerging technologies. One use these new emerging technologies. One of the things that we want to do as an of the things that we want to do as an of the things that we want to do as an organization and what we want to bring organization and what we want to bring organization and what we want to bring to the field is figure out for ourselves to the field is figure out for ourselves to the field is figure out for ourselves the best way to use AI internally in a the best way to use AI internally in a the best way to use AI internally in a responsible way in a climate informed responsible way in a climate informed responsible way in a climate informed way hopefully and then to be able to way hopefully and then to be able to way hopefully and then to be able to share those best practices with others share those best practices with others share those best practices with others in the field. So there's a huge amount in the field. So there's a huge amount in the field. So there's a huge amount that AI could do. A lot of nonprofits that AI could do. A lot of nonprofits that AI could do. A lot of nonprofits really struggle with the amount of text really struggle with the amount of text really struggle with the amount of text that they have to analyze and generate that they have to analyze and generate that they have to analyze and generate in order to apply for grants to in order to apply for grants to in order to apply for grants to understand the latest policy updates. understand the latest policy updates. understand the latest policy updates. There are all sorts of things that AI There are all sorts of things that AI There are all sorts of things that AI could help with. And so our focus right could help with. And so our focus right could help with. And so our focus right now as an as an organization is to now as an as an organization is to now as an as an organization is to figure out our best practices first and figure out our best practices first and figure out our best practices first and then open out a conversation with the then open out a conversation with the then open out a conversation with the rest of the field to try to try and help rest of the field to try to try and help rest of the field to try to try and help them too. them too. them too. >> I mean so uh so I suppose my final >> I mean so uh so I suppose my final >> I mean so uh so I suppose my final question to to uh pick your brain a bit question to to uh pick your brain a bit question to to uh pick your brain a bit is here in the context of like the is here in the context of like the is here in the context of like the Israeli high-tech sector what uh what Israeli high-tech sector what uh what Israeli high-tech sector what uh what your kind of final thoughts are on on your kind of final thoughts are on on your kind of final thoughts are on on doing AI responsibly.

doing AI responsibly. doing AI responsibly. >> Yeah, for sure. I think so AI I think a >> Yeah, for sure. I think so AI I think a >> Yeah, for sure. I think so AI I think a little bit like social media and the little bit like social media and the little bit like social media and the internet before has had an amazing internet before has had an amazing internet before has had an amazing breakout period, huge uptake, huge breakout period, huge uptake, huge breakout period, huge uptake, huge excitement about what the implications excitement about what the implications excitement about what the implications could be. Um, and I think that like the could be. Um, and I think that like the could be. Um, and I think that like the internet and like social media, we're at internet and like social media, we're at internet and like social media, we're at an inflection point now where we're an inflection point now where we're an inflection point now where we're starting to understand the nuances, the starting to understand the nuances, the starting to understand the nuances, the the dangers, the potential complexities. the dangers, the potential complexities. the dangers, the potential complexities. And I think there's also there's a And I think there's also there's a And I think there's also there's a challenge number one to develop AI challenge number one to develop AI challenge number one to develop AI responsibly and to have in mind all of responsibly and to have in mind all of responsibly and to have in mind all of those external issues, not just the those external issues, not just the those external issues, not just the bottom line, but copyright, right? bottom line, but copyright, right? bottom line, but copyright, right? Climate, um, the dangers of AI, Climate, um, the dangers of AI, Climate, um, the dangers of AI, chatbots, telling people what to do and chatbots, telling people what to do and chatbots, telling people what to do and befriending people instead of them being befriending people instead of them being befriending people instead of them being outside in the real world. All these outside in the real world. All these outside in the real world. All these things are massive challenges. And I

things are massive challenges. And I things are massive challenges. And I think that also lights the way for where think that also lights the way for where think that also lights the way for where people can be innovative and really lead people can be innovative and really lead people can be innovative and really lead the market because we're now I think the market because we're now I think the market because we're now I think moving into a second stage where the AI moving into a second stage where the AI moving into a second stage where the AI innovations aren't just how good is your innovations aren't just how good is your innovations aren't just how good is your model, how much sort of how many model, how much sort of how many model, how much sort of how many parameters can it crunch, but are you parameters can it crunch, but are you parameters can it crunch, but are you developing an application that's energy developing an application that's energy developing an application that's energy efficient and that's socially efficient and that's socially efficient and that's socially responsible. responsible. responsible. >> Amazing. Well, definitely an important >> Amazing. Well, definitely an important >> Amazing. Well, definitely an important conversation and one that's only going conversation and one that's only going conversation and one that's only going to get more important. Uh Joe, thank you to get more important. Uh Joe, thank you to get more important. Uh Joe, thank you so much for your time. so much for your time. so much for your time. >> Thank you for having me. >> Thank you for having me. >> Thank you for having me. >> Pleasure. And enjoy the rest of uh the >> Pleasure. And enjoy the rest of uh the >> Pleasure. And enjoy the rest of uh the conference. conference. conference. >> Thank you so much. Thank you.