Energy to train AI tools, wasted?

Energy used to make and provide online services is an important consideration for many reasons. Production and delivery of online services require energy, and that energy has a direct impact on the environment. The energy used to create and provide online services often comes from burning fossil fuels, such as coal, natural gas, and oil. This burning releases carbon dioxide (CO2) and other pollutants into the atmosphere, contributing to global warming. Burning fossil fuels also releases other harmful pollutants, such as particulate matter, sulfur dioxide, and nitrogen oxides, contributing to air pollution and can cause serious health problems. Increased energy consumption also has a direct effect on our environment. As energy consumption increases, so does the demand for resources such as coal, natural gas, and oil. This can lead to the destruction of ecosystems and habitats, as well as the displacement of communities. Additionally, burning these resources to produce energy contributes to climate change, causing a shift in weather patterns, rising sea levels, and an increase in extreme weather events. The energy used to provide online services also has an impact on the cost of providing these services. The more energy used to power the servers and networks, the more expensive the services become. Additionally, higher energy costs can lead to higher consumer prices, as companies must pass on the extra costs to their customers. Finally, suppose energy used to provide online services is generated from non-renewable sources, such as coal and oil. In that case, it means that the energy used to power these online services will eventually run out, which could negatively impact the availability of these services in the future. Overall, it is essential to consider the energy used to make and provide online services. Burning fossil fuels to power these services contributes to air pollution and global warming while also increasing costs. Additionally, the use of non-renewable resources to generate energy could lead to a decrease in the availability of these services in the future. 

A prominent example of online services is AI chatbots that can provide the user with answers to almost any topic. Other than a search engine that only finds matches of the search text in the indexed documents, AI chatbots can compose new information by drawing connections between the vast amount of information they have been trained with. AI programs like ChatGPT are a highly relevant development because they significantly improve the user experience and enable people from all domains to access sophisticated AI technology. Open AI programs make AI more accessible, allowing developers to share and collaborate on AI models. It also helps reduce development costs and makes integrating AI into existing applications easier. By allowing developers to access and build upon existing models, they can create new and innovative applications that can benefit everyone. Developing AI models helps automate tedious tasks and reduce the time spent on manual labor. By using AI models, businesses can automate mundane tasks and improve their workflow. AI models can also help to improve customer support and increase customer satisfaction. AI models are also important for predicting future trends and predicting customer behavior.

But, despite the fact that users of AI often get free access or a generous free trial, developing and training an AI model does not come for free when we consider the energy budget. The Carbon footprint of training ChatGPT has been estimated to be 1287 MWh [1], in addition to running the services. Are 1287 MWh a number to be concerned with? Probably yes. Is it a number so high that we immediately need to banish AI training for the sake of the environment? I don't think so.

When relating 1287 MWh to a single person, it is a lot. It would mean driving an average European car on fossil fuels for 4,5 Mio km. That is enough to travel the whole road network of the United States or equivalent to the carbon footprint of a flight passenger going form London to New York 320 times.

Nevertheless, ChatGPT has more than one user. In fact, it is one of the fastest-growing online platforms in the world, with around 100 Million users at the time of writing. Dividing the development costs by the users, it amounts to 0.01287 kWh or roughly 1% of the energy required to print a book. 

In other words, if users can utilize the AI system to automate mundane tasks and improve their workflow, the energy spent on creating the AI is probably well-invested. Many usages are recreational, and sometimes the AI provides more fiction than facts, but so is the case with books.

However, we need to keep our eyes open on two issues:

• The operational cost of running the system: "Cost" would mean here energy cost as well as financial cost. If the system does not work here efficiently, we could end up in a much higher energy waste than 1287 MWh
• Further developments in training new AIs: competitors might train their own AIs, no matter the (energy) cost. And models are expected to grow in complexity and capabilities, probably also significantly raising the energy required for training a single model.

So let's keep an eye on further developments.

[1] Patterson, D., Gonzalez, J., Hölzle, U., Le, Q., Liang, C., Munguia, L.-M., … Dean, J. (4 2022). The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink. Computer, 55, 18–28. Retrieved from http://arxiv.org/abs/2204.05149

GREEND: An Energy Consumption Dataset of Households in Italy and Austria

Some days ago, we got notice that our paper "GREEND: An Energy Consumption Dataset of Households in Italy and Austria" is accepted at the IEEE SmartGridComm 2014. The SmartGridComm 2014 will take place in Venice, a fascinating and marvelous Italian city, on November 3-6, 2014.

GREEND: An Energy Consumption Dataset of Households in Italy and Austria, 
Andrea Monacchi, Dominik Egarter, Wilfried Elmenreich, Salvatore D’Alessandro, Andrea M. Tonello

Home energy management systems can be used to monitor and optimize consumption and local production from renewable energy. To assess solutions before their deployment, researchers and designers of those systems demand for energy consumption datasets. In this paper, we present the GREEND dataset, containing detailed power usage information obtained through a measurement campaign in households in Austria and Italy. We provide a description of consumption scenarios and discuss design choices for the sensing infrastructure. Finally, we bench- mark the dataset with state-of-the-art techniques in load disaggregation, occupancy detection and appliance usage mining.

An authors' version is available in arXiv: http://arxiv.org/abs/1405.3100

The GREEND dataset is publicly availble, donwload the newest version at https://sourceforge.net/projects/greend/files/GREEND_0-2_300615.zip

How do you use your electrictiy?

MONERGY

The MONERGY project aims to develop innovations that contribute to a more efficient energy consumption in households. We look specifically at the situation in the regions of Carinthia and Friuli-Venezia Giulia.

To guide our research efforts, we need your help in the form of a short survey (approximately 10 minutes) on the use of electrical appliances at home:

• Survey: Energieverbrauch (in German language)
• Survey: Consumo di energia (in Italian language)

Since we especially aim at the situation in Carinthia and Friuli-Venezia Giulia, the survey is only available in Italian and German. If you don't live in that region, your contribution is still helpful and appreciated, but there is no English version, sorry.

The results and conclusions will be published later on via the MONERGY website (in German/Italian) and in this blog (in English language).

Looking forward to your participation in the survey!

Energy Aware Software-Engineering and Development (EASED@BUIS)

2. Workshop on Energy Aware Software-Engineering and Development (EASED@BUIS)

25.04.2013

Oldenburg (Oldb.)
Kulturzentrum PFL
Peterstraße 3

The EASED workshop will be held in conjunction with the 5th BUIS-Days: IT-based resource and energy management.
Utilization of mobile and embedded devices, and thus their induced energy consumption, is constantly increasing. Reducing the energy consumption of such devices will not only improve the carbon footprint of contemporary mobile IT usage, but will also extend the device lifetime, improve user acceptance and reduce operational costs.
Next to serious and ongoing efforts in hardware design and on operating system level, software engineering techniques will also contribute to optimize energy consumption by improving software design and software quality. The EASED@BUIS workshop, which follows up the Workshop on Developing Energy Aware Software Systems (EEbS 2012), held at the annual GI Conference in September 2012, provides a broad forum for researchers and practitioners to discuss ongoing works, latest results, and common topics of interest regarding the improvement of software induced energy consumption.
Intensive discussions at the first workshop identified a major challenge in optimizing energy efficiency, which is to precisely measure energy consumption of software regarding user behavior.
Thus, the follow workshop EASED@BUIS will focus on the following topics:

• approaches and techniques to estimate or measure the energy consumption of software components,
• approaches to define standardized usage scenarios of applications on mobile devices to provide repeatable measurement of energy consumption in concrete application settings,
• approaches to model the energy consumption of software components, and
• experiences on measuring and improving the energy consumption of software components. 

Well elaborated and standardized measurement means will provide an important foundation to detecting sources of wasting energy caused by software systems and will enable validation means to verify energy savings by software improvements.
EASED@BUIS will be organized as a one day discussion-intensive workshop to provoke intensive collaborations among the participants. It is intended to initiate collaborative works on standardizing (static and dynamic) measuring techniques for energy consumption.
To further stimulate these discussions, authors are invited to submit position papers one on the        workshop´s topics. Accepted papers will be presented at the workshop and will be published in Softwaretechnik-Trends.
EASED@BUIS is supported by the GI special Interest groups Software Technology and Environmental Informatics.

Submissions and important Dates
Authors are encouraged to submit their position papers (2 pages in two column form) not later than March 15, 2013 through easychair.
paper submission deadline: March 15, 2013
author notification: March 25, 2013
camera-ready deadline: April 1, 2013
Organizing Committee
Christian Bunse (University of the Applied Sciences Stralsund)

Stefan Naumann (University of the Applied Sciences Trier, Environmental Campus Birkenfeld)

Andreas Winter (Carl von Ossietzky University, Oldenburg)

Program Committee
Colin Atkinson (University Mannheim)
Paris Avgeriou (University of Groningen)
Holger Eichelberger (University Hildesheim)
Sebastian Götz (TU Dresden)
Theo Härder (TU Kaiserslautern)
Mirco Josefiok (OFFIS, Oldenburg)
Ákos Kiss (University of Szeged)
Sonja Klingert (University Mannheim)
Patricia Lago (VU University Amsterdam)
Thierry Leboucq (KaliTerre, Nantes)
Birgit Penzenstadler (TU München)
Giuseppe Scanniello (University of Basilicata)
Maximilian Schirmer (Bauhaus-University Weimar)
Gunnar Schomaker (OFFIS Oldenburg)
Joost Visser (Software Improvement Group, Amsterdam)
Claas Wilke (TU Dresden)
Alexandru Telea (University of Groningen)
Local Organization
Marion Gottschalk (Carl von Ossietzky University, Oldenburg)
Andreas Winter (Carl von Ossietzky University, Oldenburg)

Evolving Non-Intrusive Load Monitoring

Our paper Evolving Non-Intrusive Load Monitoring by Dominik Egarter, Anita Sobe and Wilfried Elmenreich has been accepted for the conference track EvoEnergy (Evolutionary Algorithms in Energy Applications) of the EvoApplication (16th European Conference on the Applications of Evolutionary Computation) 2013, taking place in Vienna form 3rd to 5th of April.

Basic principle of the ON/OFF time genome appliance
detection. Given is the total power consumption over
time. The goal is to deduce the on/off times of devices
(colored blocks) that add up to the measured power profile.

Non-intrusive load monitoring (NILM) identifies used appliances in a total power load according to their individual load characteristics. In this paper we propose an evolutionary optimization algorithm to identify appliances, which are modeled as on/off appliances. We evaluate our proposed evolutionary optimization by simulation with Matlab, where we use a random total load and randomly generated power profiles to make a statement of the applicability of the evolutionary algorithm as optimization technique for NILM. Our results shows that the evolutionary approach is feasible to be used in NILM systems and can reach satisfying detection probabilities.



Dominik Egarter, Anita Sobe, Wilfried Elmenreich,   Evolving Non-Intrusive Load Monitoring,   EvoApplication 2013,   16th European Conference on the Applications of Evolutionary Computation, April, 2013

Workshop on Modeling and Simulation of Cyber-Physical Energy Systems 2013

Workshop on Modeling and Simulation of Cyber-Physical Energy Systems 2013

May 20 2013, Berkeley CA

http://www.palensky.org/mscpes/

Modern energy systems combine information technology, electrical and thermal infrastructure, autonomous roles and interact with other systems like markets and regulations. Existing modeling and simulation tools are not capable to cover such systems in all of their aspects, new languages, methods and tools are necessary. A combination of universal modeling languages like Modelica and established, specialized tools like grid simulators and telecommunication simulators is necessary. This leads to modeling and co-simulating hybrid systems where for instance a multi-agent framework and an electric grid simulator are combined to investigate smart electric vehicle charging algorithms. It is especially the potential size of such systems that constitute a challenge for modeling and simulation. Implementing these future CPS are another substantial challenge. The designed algorithms need to be compact, computationally inexpensive, potentially self-organizing and intrinsically stable if applied to real energy systems. New methods and alternative ways are necessary to overcome these challenges.

This workshop is a platform for researchers and developers to exchange ideas to the following (not exhaustive) list of topics:

-    Hybrid modeling and simulation

-    Co-Simulation

-    High-performance computing

-    Analytics of system data

-    Ontologies for energy systems

-    Applications of cyber-physical energy systems

-    Distributed algorithms and control

-    Standards in interfacing components

-    Numerics for hybrid and co-simulation

-    Formal languages for energy systems

-    Smart Grid modeling

-    Demand response and power quality

-    Information and communication technology for intelligent energy systems

Submitted papers are peer-reviewed by at least 3 reviewers. Workshop language is English. Proceedings will be published by Springer.

Full paper submission: January 31, 2013

Notification of acceptance: February 20, 2013

Your workshop chairs,

Peter Palensky (Austrian Institute of Technology)

Edward A.Lee (University of California Berkeley)

Home appliance energy usage

The Smart Grid will help to balance energy production and consumption.
While we are waiting for the Smart Grid to come, there is something we can do meanwhile - optimizing the energy usage of our local network of electrical appliances. Because one thing is for sure: energy does not come for free now and won't come for free in the future either.
When doing optimization, the important thing is to identify the parts of a system, where an optimization significantly affects the overall outcome. In the blog of the General Electric Company, you can find a nice interactive visualization of the power consumption for a selectable set of typical appliances.

A zero-Watt cloth dryer
(source:Wikimedia commons)

The website application tries to guess power consumption and cost of your set of appliances. While the values are just rough estimates, the numbers still give you a feeling for appliances which are expensive in terms of energy consuming and appliances which are more frugal.

I personally was quite surprised by the high power consumption of an electrical cloth dryer. Luckily, I have the zero-Watt edition at home :-)