Get started

Emissions

The environmental impact of AI: a case studyIn our previous blog, Will AI workloads consume all the world’s energy?, we looked at the relationship between increasing processing power and an increase in energy demand, and what this means for artificial intelligence (AI) from an environmental standpoint. In this latest blog, we aim to further illuminate this discussion with a case study of the world’s biggest large language model (LLM), BLOOM. Case study on environmental impact: BLOOM An accurate estimate of the environmental impact of an LLM being run is far from a simple exercise. One must understand, first, that there is a general ‘model life cycle.’ Broadly, the model life cycle could be thought of as three phases1: Inference: This is the phase when a given model is said to be ‘up-and-running.’ If one is thinking of Google’s machine translation system, for example, inference is happening when the system is providing translations for users. The energy usage for any single request is small, but if the overall system is processing 100 billion words per day, the overall energy usage could still be quite large. Training: This is the phase when the parameters of a model have been set and the system is exposed to data from which it is able to learn such that outputs in the inference phase are judged to be ‘accurate’. There are cases where the greenhouse gas emissions impact for training large, cutting-edge models can be comparable to the lifetime emissions of a car. Model development: This is the phase when developers and researchers are seeking to build the model and will tend to experiment with all sorts of different options. It is easier to measure the impact of training a finished model that becomes public, as opposed to seeking to measure the impact of the research and development process, which might have included many different paths prior to getting to the finished model that the public actually sees. Therefore, the BLOOM case study focuses on the impact from training the model. BLOOM is trained on 1.6 terabytes of data in 46 natural languages and 13 programming languages. Note, at the time of the study, Nvidia did not disclose the carbon intensity of this specific chip, so the researchers needed to compile data from a close approximate equivalent setup. It’s an important detail to keep in mind, in that an accurate depiction of the carbon impact of training a single model requires a lot of information and, if certain data along the way is not disclosed, there must be more and more estimates and approximations (which will impact the final data). If AI workloads are always increasing, does that mean carbon emissions are also always increasing2? Considering all data centres, data transmission networks, and connected devices, it is estimated that there were about 700 million tonnes of carbon dioxide equivalent in 2020, roughly 1.4% of global emissions. About two-thirds of the emissions came from operational energy use. Even if 1.4% is not yet a significant number relative to the world’s total, growth in this area can be fast. Currently, it is not possible to know exactly how much of this 700 million tonne total comes directly from AI and machine learning. One possible assumption to make, to come to a figure, is that AI and machine learning workloads were occurring almost entirely in hyperscale data centres. These specific data centres contributed roughly 0.1% to 0.2% of greenhouse gas emissions. Some of the world’s largest firms directly disclose certain statistics to show that they are environmentally conscious. Meta Platforms represents a case in point. If we consider its specific activities: Overall data centre energy use was increasing 40% per year from 2016. Overall training activity in machine learning was growing roughly 150% per year. Overall inference activity was growing 105% per year. But Meta Platforms’ overall greenhouse gas emissions footprint was down 90% from 2016 due to its renewable energy purchases. The bottom line is, if companies just increased their compute usage to develop, train and run models—increasing these activities all the time—then it would make sense to surmise that their greenhouse gas emissions would always be rising. However, the world’s biggest companies want to be seen as ‘environmentally conscious’, and they frequently buy renewable energy and even carbon credits. This makes the total picture less clear; whilst there is more AI and it may be more energy intensive in certain respects, if more and more of the energy is coming from renewable sources, then the environmental impact may not increase at anywhere near the same rate. Conclusion—a fruitful area for ongoing analysis One of the interesting areas for future analysis will be to gauge the impact of internet search with generative AI versus the current, more standard search process. There are estimates that the carbon footprint of generative AI search could be four or five times higher, but looking solely at this one datapoint could be misleading. For instance, if generative AI search actually saves time or reduces the overall number of searches, in the long run, more efficient generative AI search may help the picture more than it hurts3. Just as we are currently learning how and where generative AI will help businesses, we are constantly learning more about the environmental impacts. Sources 1 Source: Kaack et al. “Aligning artificial intelligence with climate change mitigation.” Nature Climate Change. Volume 12, June 2022. 2 Source: Kaack et al., June 2022. 3 Source: Saenko, Kate. “Is generative AI bad for the environment? A computer scientist explains the carbon footprint of ChatGPT and its cousins.” The Conversation. 23 May 2023.
SPARKS:AIX
by aneekaguptaWTE
Will AI workloads consume all the world’s energy?On big questions like this, almost nothing stays constant. When we consider a new technology: We cannot assume that rates of adoption or usage will remain constant—they may drop, they may even grow. We cannot assume that the technology supplying our energy needs will remain constant—there could be breakthroughs in efficiency or changes in the overall energy mix. We cannot assume that the efficiency of the specific technology being adopted will remain constant—we have seen numerous examples of areas where an initial version of something in technology or software faces subsequent improvements that may give it greater capabilities with lower energy usage. We must also recognise that artificial intelligence (AI) itself could suggest improvements in energy efficiency for specific applications—like the heating and cooling of a building. Therefore, any analysis of energy usage and AI must recognise that the one constant will be change. Environmental impact of select large language models (LLMs) LLMs have been garnering the lion’s share of attention amidst the current excitement around generative AI. It makes sense to consider the amount of carbon emissions generated by some of these systems. The Stanford AI Index Report, published in 2023, provided some data, noting that factors like the number of parameters in a model, the power usage effectiveness1 of a data centre, and the grid carbon intensity all matter. Considering power consumption of an LLM Those building different LLMs have many levers they can pull in order to influence different characteristics, like energy consumption. Google researchers proposed a family of language models named GLaM (Generalist Language Model), which uses a ‘sparsely activated mixture of experts’. While a full discussion of how that type of approach works is beyond the scope of this piece, we note that the largest of the GLaM models has 1.2 trillion parameters. Knowing solely that data point, the assumption would be that this model would consume more energy than any of the models. In reality, the GLaM model with 1.2 trillion parameters consumes only one-third of the energy required to train GPT-3 and requires only half of the computation flops for inference operations. A simple way to think of what is going on is that, while the total model has 1.2 trillion parameters, a given input token into the GLaM model is only activating a maximum of 95 billion parameters, that is, the entire model isn’t active across all the parameters. GPT-3, on the other hand, activated all 175 billion parameters on each input token3. It is notable that, even if measuring the performance of AI models occurs on many dimensions, by many measures the GLaM model is able to outperform GPT-3 as well4. Conclusion The bottom line is that model design matters, and if model designers want to denote ways to maintain performance but use less energy, they have many options. Sources 1 Power usage effectiveness (PUE) is useful in evaluating the energy efficiency of data centres in a standard way. PUE = (total amount of energy used by a computer data centre facility) / (energy delivered to computer equipment). A higher PUE means that the data centre is less efficient. 2 Source: Du et al. “GLaM: Efficient Scaling of Language Models with Mixture-of-Experts.” ARXIV.org. 1 August 2022. 3 Source: Patterson, David; Gonzalez, Joseph; Hölzle, Urs; Le, Quoc Hung; Liang, Chen; Munguia, Lluis-Miquel; et al. (2022): The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink. TechRxiv. 4 Source: Du et al, 1 August 2022.
Editors' picks
SPARKS:AIX
by aneekaguptaWTE
9
Schneider (SU.pa) bullish scenario:The technical figure Triangle can be found in the daily chart in the French company Schneider Electric (SU.pa). Schneider Electric SE is a French multinational company that specializes in digital automation and energy management. It addresses homes, buildings, data centers, infrastructure and industries, by combining energy technologies, real-time automation, software, and services. The Triangle broke through the resistance line on 19/04/2023. If the price holds above this level, you can have a possible bullish price movement with a forecast for the next 9 days towards 155.86 EUR. According to experts, your stop-loss order should be placed at 141.38 EUR if you decide to enter this position. Schneider Electric S.E. (SU.PA)’s target to achieve net-zero operational emissions and to reduce Scope 3 emissions by 35% by 2030 (compared to Scope 3 emissions in 2017) was validated by the SBTi in 2019. Schneider's EcoStruxure solutions assisted customers in reducing carbon emissions by 84 million tonnes in 2021, totaling 347 million tonnes saved or avoided since 2018. Moreover, since launching The Zero Carbon Project in April 2021, Schneider Electric S.E. (SU.PA) has been working closely with 1,000 of its top suppliers to reduce operational carbon emissions by half by 2025. Risk Disclosure: Trading Foreign Exchange (Forex) and Contracts of Difference (CFD's) carries a high level of risk. By registering and signing up, any client affirms their understanding of their own personal accountability for all transactions performed within their account and recognizes the risks associated with trading on such markets and on such sites. Furthermore, one understands that the company carries zero influence over transactions, markets, and trading signals and cannot be held liable nor guarantee any profits or losses.
EURONEXT:SULong
by legacyFXofficial
WKHS Swing-Trade IdeaWKHS (Workhorse Group, Inc.) is a swing-trade idea that presents the following (Bullish) conditions: 1.) The 5-day moving average is about to cross the 21-day moving average price. When short period moving averages cross above long period moving averages, it tends to indicate an upward price movement. 2.) The monthly "Moving Average Convergence/Divergence" (MACD) histogram (indicator at bottom) is ending a down wave and beginning a new up wave. 3.) The concentration of volume (Blue/Yellow bars on left) indicates that the majority of investors currently own stock at the $16.20 price level (Point of Control), this level drops by approximately half as the price moves up towards the $17.92 (Yearly) Fibonacci support/resistance level. The volume of of share holders above $17.92 decreases until about $19.57 (Less resistance). Then resistance increases until $21.37, followed by another decrease in resistance until the price hits the next fibonacci support/resistance level at $22.71. This is followed by decreasing concentrations of volume as the price moves up. For movement up beyond this point the following will be critical. 4.) Workhorse was the 2nd place bidder on A major contract to update the postal delivery vehicle fleet for the United States Postal Service (USPS). Workhorse is offering a 100% electric vehicle fleet. The 1st place bidder OSK (Oshkosh, Corporation.) was awarded the contract by the outgoing Trump appointed head of the USPS. Oshkosh is only offering 10% of the fleet as electric vehicles and 90% internal combustion engine. The House of Congress and the Biden Administration may rescind the awarded contract to OSK to meet the requirements of an Executive order that President Biden signed on 25/Jan/2021. If Workhorse is granted the contract, look for a move up from the $22.71 to $27.49 fibonacci support/resistance levels. Just my opinion, not advice. Thank you.
NASDAQ:WKHSLong
by khollspati
Speculating on global warming amount, and future trendHello, in this pseudo-science idea I try to get a vague idea of how unrealistic is thinking earth will just turn into a fireball, and quantify all this. It's all unprecise and no idea what the real numbers are but I'll work with worse case scenarios and min max, as to get an idea of what order of magnitude to expect. The idea is really not to get a precise estimate but an idea of the possible MIN and MAX. I'll assume the following: Between AD 1000 and 1800 CO2 atmospheric levels were around 280, and from 1800 to 2000 they rose to 400, let's assume all 120 was manmade (past 10,000 years levels have been very slowly going up so it makes sense to believe it was mostly manmade). We have enough fossil fuels to burn to raise earth atmosphere levels by 1000 ppm including idk 200 from melting ice. I haven't been able to find how much co2 would melting all ice release (what a surprise) and neither a good estimate of how much burning all reserves would do (you'd think they would bother looking at this if their lives were threatened). But considering 6.66 times what we emitted already since 1800 I think is fair. Some data: Earth's atmosphere contains 3,200,000,000,000 (3.2 trillion) tonnes of CO2 (0.04%). Earth mass = 5.972 × 10²⁴ kg. The average temperature on the Moon (at the equator and mid latitudes) varies from -298 degrees Fahrenheit (-183 degrees Celsius), at night, to 224 degrees Fahrenheit (106 degrees Celsius) during the day. No atmosphere there (10 metric tonnes...). Moon mass = 7.347 x 10²² kg (1.23% earth). Water is earth bigger warming contributor. When CO2 goes up, plants may be able to take more H20 in, also NASA has observed earth and it is greener. So, when CO2 goes up, water, the top global warming gaz, gets sucked up from the atmosphere. No idea how big of a difference this makes. Temperatures: Earth 289°K Venus 743°K Mercury 700 degrees Kelvin in the day, minus 93 K at night. Average temperature of 440 K. Mars 213 deg K or 218K??? Moon 379°K at day, 90°K at night. Pressures: Earth 1- Considering there is a direct correlation between CO2 quantity & temperature. a- Compared to Mars Mars is 11% the size of earth, and 95% of its atmosphere is CO2. There is 23,750,000,000,000 (23.75 trillion) tonnes. +6 degrees assuming all of those 6 degrees are cause by CO2, means an increase of 0.25263°K per trillion tonne of CO2. Also we assume earth has the same correlation. So say you increase earth CO2 up to 1400 ppm. The quantity of CO2 goes from 3.2 trillion tonnes to 11.2, or 8*10^12 tonnes are added. ==> +2°K/°C or + 3.6°F. An increase of 120 ppm using this formula would cause + 0.2425°C or 0.4365°F. Since industrial age temperature went up 0.7°C if I recall. So it seems plausible that a third of it was due to human activity (and 2/3 because of natural activities). Not sure how much it went up since the end of the little ice age in 1850. Mercury has +4 degrees. What if we assume Mars has the same? And so then CO2 only amounts to + 2 degrees? Then: + 1000 ppm in earth atmosphere ==> +0.666°K/°C or + 1.2°F Since the industrial age ==> +0.08°K/°C or 0.144°F Which seems plausible and reasonable. b- Compared to Venus Venus has ridiculously high levels of CO2. +503 degrees (K/C) for 460 million trillion tonnes of CO2. H20 is too small to be relevant here. So same, we just assume direct correlation. For every trillion tonne of CO2 added, temperature goes up 1,093478e-6 (0,000001093478) °K. + 1000 ppm ==> +8 trillion tonnes = 0,00000875 degrees Since the industrial age ==> + ~ 0,000001 degrees 2- And I won't go further but we could include planet size, atm pressure, other factors... For example, since mars is much smaller than earth, one could assume that 1 tonne of CO2 has a greater effect on Mars than on Earth. 3- What about comparing to earth? If we assume all warming since the little ice age was man made? First I doubt this is true. Temperatures were in the low area of support historically. And it started going up before emissions. But say we assume 0.8°K were the cause of human activities. In the 1950/1960 to 2000 period, when harmful chemicals were being released in the atmosphere (CFCs etc), temperature went up about 0.65°K. So outside of this we got a +0.2°K in 100 years? And temperature has flattened or barely went up since 2000. Well that depends how "adjusted" your data is. So without CFCs what? +0.25°K for a 120 ppm increase? It's all speculation, this is so unscientific. So at most +2.08°K for a 1000 ppm increase. This is consistent with the estimate using mars. For me, the absolute max, if all of earth warming was manmade is 2 degrees for an increase of 1000 ppm (8.33 what man has emitted until now). How much can CO2 concentrations go up realistically? Between 2000 and 2020 the level went up from 370 to 410, so +40. Between 1980 and 2000 the level went up from 340 to 370, so +30. Between 1960 and 1980 the level went up from 320 to 340, so +20. USA emissions have peaked in 2000 or the early 2000s and is declining. China peaked if I recall. Europe peaked. Then the big ones are India and Africa. Well anyway, let's say it keeps going up a bit then peaks at double what is is now, 80 every 20 years. Let's say for the next 100 years this is what we get. 5 * 80 = 400 ppm. This would lead to an increase of 400 ppm. Maybe a bit more with ice melting, but this won't be hundreds. We probably will get at most half of the 1000 I used in my examples. So if I were to bet money, I would not bet on an average of more than 1 Kelvin for the next 100 years. At the very most, but probably under that. The effect of CO2 on °K has to be more complex than a simple linear correlation, and there has to be diminishing returns. It is a shame we don't have historical water contents, not that I know of. All I know is that CO2 and H20 were super high billions of years ago when life appeared. But anyway, that 0.00014% to 33% of the rise in temperatures since the end of the little ice age can be attributed to the increase in atmospheric CO2 seems reasonable. 1 to 10% seems the most reasonable but this isn't a fact. Also there is the small detail that earth temperature went up sharply exactly as Chlorofluorocarbons levels went up, and after their levels topped in 1990, earth temperature topped... Tiny irrelevant detail I know. Here are all the greenhouse gases concentrations (except water): cdiac.ess-dive.lbl.gov Methane is pretty annoying I don't see how we could stop this one without all starving. Red meat is a big problem, and for some reason people are obssessed with red meat. We can't increase methane levels tremendously forever. I'm not too worried about CO2, we'll run up of fossil fuels eventually, and raising the levels a bit helps plants grow, I just don't see how bad it can be. CFCs and other crap (Hydrofluorocarbon-23 (CHF3), Sulphur hexafluoride (SF6), PFC-14 (CF4)) have thousands of times the global warming potential CO2 has even according to "the establishment" that hates CO2, and stay in the atmosphere for millenias. BUT we finally stopped trolling and polluting the planet with this crap. That was really insane. Methane thought... that one could be a big nuisance. Agriculture is releasing levels so huge. It disappears fast but does it just turn to CO2? If it peaks at a few thousands parts per billion, that's only a few ppm, and this disappears in 25 years, it would not add much CO2. A constant methane level that does some warming and then a tiny increase of CO2 level, maybe that's not that scary. Over the long term thought what would happen? All of this also agrees with the global warming going on Mars. With all the crazies that think they are going to die you'd think we'd know more on the subject... www.nasa.gov You also got Pluto that is warming while it distances itself from the sun. www.newscientist.com But it could be a coincidence, that thing alarmists deny exists. Correlation does not imply causation, unless it fits your agenda. Also, there could be a snowball effect with CO2 increasing water level in the atmosphere, but if this was the case we can all be absolutely certain we would know about it. We would not hear the end of it. There is either no increase in water levels, or they are even diminishing. If it never gets mentionned there is zero data than it is that it does not fit their agenda. Ok I found something about humidity: www.climate4you.com Surface humidity stayed flat. It's tiring to have to fight throught tons of idiotic nonsense and fear mongering and half truths to get any crumbs of data. High up it has been flat or slightly downtrending. Actually went up a little when temperature did not. And down or flat when temperature went up. I figured CFCs caused the big uptrend in earth temperature from 1950 to 2000 but I actually found a paper claiming CFCs caused global warming? Weird I never heard of this... Censorship I guess. Haven't read it yet. phys.org So I guess the trend will continue, at least small: Slight cooling for the next decades as cancer chemicals in the air levels decline or maybe the warming trend overtakes the cooling one in any case I don't expect any major move, better farmland yields with more CO2, better living conditions as long as fossil fuel reserves are high. I also expect more "data adjustments", temperature charts with extreme isolated points rather than year averages, still no answer as to why ocean temperature went up, and more 12 year ultimatums lmao pathetic liers. Well that's enough thinking for now. All I know is I won't invest in renewable companies for now. Electric cars? Never. Biofuels are good but it's 50-100 years early. I really love the idea of hydroplants also.
NYMEX:CL1!Education
by MrRenev
3