The Climate Crisis & Why Data Must Not Become the New Oil

The Footprint of Artificial Intelligence and Data Storage

Guest post by Alex Moltzau

A recent study from University of Massachusetts Amherst (Strubell et al., 2019) has shown that deep learning has a terrible carbon footprint. As an example training a single AI model can emit as much carbon as five cars in their lifetimes. We can, therefore, with good reason be highly critical of energy requirements of applied AI.

Taking this into consideration the intense applications within new army intelligence and surveillance systems could be a possible drain on the resources of our planet. The US Army bought an intelligence system from the private company Palantir (Harris, 2019); $1 billion dollars is allocated to AI for 2020 in the US Department of Defense (Breaking Defense, 2019); and the US development of industry standards for AI has no mention of climate change or carbon (NIST, 2019). This does not bode well considering the climate crisis that we are currently in.

Although techniques within the field of AI can be applied in renewable energy solutions (Kalogirou, 2011; Jha et al., 2017) it is a meagre consolation. Indeed we must consider the combined emission of large technology companies: Amazon, Apple, Microsoft, Google (Bernal & Hoggins, 2019) and beyond. It is estimated that the billions of internet-connected devices could produce 3.5% of global emissions and consume 20% of electricity by 2025 (Vidal, 2017).

In terms of the current energy mix that these large technology companies have, you might want to carefully consider which company to choose.

Which of the large technology platforms would you choose?

Amazon-AWS

  • 17% renewable,
  • 24% gas,
  • 30% coal, and
  • 26% nuclear power.

Microsoft

  • 32% renewable
  • 23% gas,
  • 31% coal and
  • 10% nuclear power.

Google

  • 56% renewable,
  • 14% gas,
  • 15% coal, and
  • 10% nuclear power.

These numbers are made according to a report published by Greenpeace (Cook, 2017). It is said by a prominent writer on AI:

The artificial-intelligence industry is often compared to the oil industry: once mined and refined, data, like oil, can be a highly lucrative commodity. Now it seems the metaphor may extend even further. Like its fossil-fuel counterpart, the process of deep learning has an outsize environmental impact.”
— Karen Hao, MIT Technology Review (Karen Hao, 2019)

Protests in Large Technology Companies

Data is a question of power and this is not new (Bridle, 2018; Susskind, 2018). Yet it is increasingly a question of responsibility for workers. The years 2018–2019 has been marked by several large protests in technology companies. Demonstrations were held at different Google offices around the world against sexual misconduct allegations (Weaver et al., 2018). At the time more than 20’000 of Google’s 94,372 full-time and contract employees worldwide joined this joint effort (Statt, 2018) organised on the 1st of November 2018.

In 2019 employees from Amazon, Microsoft, Salesforce and other technology firms protested collaboration that these companies had with the ICE — the Immigration and Customs Enforcement (Sandler, 2019). ICE is still engaging software companies, and one employee in the software company Chef deleted his code in protest of ICE. Chef did not cancel their contract with ICE and instead republished the code (Finley, 2019).

There has additionally been protests from app-based employees regarding labour rights (Ongweso, 2019; Alva, 2019). This has become increasingly important for ride-hailing tech giants, construction, healthcare, trucking, janitorial services, nail salons, adult entertainment, commercial fishing and newspapers. State Senator Maria Elena Durazo commenting on a new bill said: “Let’s be clear: there is nothing innovative about underpaying someone for their labor,” (Conger & Scheiber, 2019). App-based technology companies have not liable to take care of their workers and this may change.

At this point, you may find yourself wondering why you are reading about protests in an article about artificial intelligence. Perhaps I want you to understand that there are issues in tech, and not everyone agrees with the way it is being handled. Amazon as an example has the largest data center capacity in the world yet is also the largest emitter (Greenpeace USA, 2019; Stoller, 2019)

Amazon.com, inc. is diversified. It is the largest e-commerce marketplace, AI assistant provider, and cloud computing platform as measured by revenue and market capitalization. Amazon is currently the largest Internet company by revenue in the world (Amazon.com, 2019). At the same time, Amazon emitted 44.4 million tonnes of carbon dioxide in 2018 (CBC, 2019).

Since artificial intelligence requires a lot of computing power and storage the activities of Amazon is therefore vital to discuss. In August 2019 there was no company-wide plan to reach zero carbon emissions and their goal to reach 100% renewable energy had no date for completion. There was a large Oil & Gas initiative devoted to helping fossil fuel companies accelerate and expand oil and gas extraction. Amazon donated to 68 members of congress in 2018 who voted against climate legislation 100% of the time (Amazon Employees for Climate Justice, 2019).

Amazon is actively fighting or lacking emissions transparency (Bogle, 2019; Roston 2019) and has many shell companies so it can be hard to track. There are efforts by researchers to reduce the carbon footprint of servers (Avgerinou et al., 2017; Berl et al., 2010; Buyya et al., 2010; Khosravi et al., 2013). However, there is far more to consider if we take a resource perspective.

References

Alva, E. (2019, September 22). For too long Lyft and Uber have abused drivers like me. Not any more | Edan Alva. Retrieved from https://www.theguardian.com/business/2019/sep/22/california-ab5-uber-lyft-drivers

Amazon.com. (2019, January 31). Amazon.com Announces Fourth Quarter Sales up 20% to $72.4 Billion: Amazon.com, Inc. Retrieved from https://ir.aboutamazon.com/news-releases/news-release-details/amazoncom-announces-fourth-quarter-sales-20-724-billion

Amazon.com Announces Fourth Quarter Sales up 20% to $72.4 Billion: Amazon.com, Inc. (2019, January 31). Retrieved from https://ir.aboutamazon.com/news-releases/news-release-details/amazoncom-announces-fourth-quarter-sales-20-724-billion

Avgerinou, M., Bertoldi, P., & Castellazzi, L. (2017). Trends in data centre energy consumption under the european code of conduct for data centre energy efficiency. Energies, 10(10), 1470.

Bernal, N., & Hoggins, T. (2019, May 3). Amazon, Apple and Google are gobbling up the world’s energy. What are they doing about it? Retrieved from https://www.telegraph.co.uk/technology/2019/05/03/amazon-apple-google-gobbling-worlds-energy/

Berl, A., Gelenbe, E., Di Girolamo, M., Giuliani, G., De Meer, H., Dang, M. Q., & Pentikousis, K. (2010). Energy-efficient cloud computing. The computer journal, 53(7), 1045–1051.

Bogle, A. (2019, July 3). Amazon doesn’t want you to know its greenhouse emissions, says they’re a ‘trade secret’. Retrieved from https://www.abc.net.au/news/science/2019-07-04/amazon-emissions-energy-data-australia-transparency-questions/11268918

Bridle, J. (2018). New dark age: Technology and the end of the future. Verso Books.

Buyya, R., Beloglazov, A., & Abawajy, J. (2010). Energy-efficient management of data center resources for cloud computing: a vision, architectural elements, and open challenges. arXiv preprint arXiv:1006.0308.

CBC. (2019, September 20). Jeff Bezos makes climate pledge, after revealing Amazon emits as much carbon as a small nation | CBC News. Retrieved from https://www.cbc.ca/news/business/amazon-climate-1.5291076

Cook, G., Lee, J., Tsai, T., Kong, A., Deans, J., Johnson, B., & Jardim, E. (2017). Clicking Clean: Who is winning the race to build a Green Internet?. Greenpeace Inc., Washington, DC, 5.

Conger, K., & Scheiber, N. (2019, September 11). California Passes Landmark Bill to Remake Gig Economy. Retrieved from https://www.nytimes.com/2019/09/11/technology/california-gig-economy-bill.html

Couts, A. (2018, June 25). Salesforce Workers Urge CEO to ‘Re-Examine’ Work With Customs and Border Protection. Retrieved from https://gizmodo.com/salesforce-workers-urge-ceo-to-re-examine-work-with-cus-1827118343

Finley, K. (2019, September 20). A Developer Deletes His Code to Protest Its Use by ICE. Retrieved from https://www.wired.com/story/developer-deletes-code-protest-ice/

Greenpeace USA. (2019, February 13). Clicking Clean Virginia. Retrieved from https://www.greenpeace.org/usa/reports/click-clean-virginia/

Hao, K. (2019, July 23). Training a single AI model can emit as much carbon as five cars in their lifetimes. Retrieved from https://www.technologyreview.com/s/613630/training-a-single-ai-model-can-emit-as-much-carbon-as-five-cars-in-their-lifetimes/

Harris, S. (2019, March 26). Palantir wins competition to build Army intelligence system. Retrieved from https://www.washingtonpost.com/world/national-security/palantir-wins-competition-to-build-army-intelligence-system/2019/03/26/c6d62bf0-3927-11e9-aaae-69364b2ed137_story.html

Jha, S. K., Bilalovic, J., Jha, A., Patel, N., & Zhang, H. (2017). Renewable energy: Present research and future scope of Artificial Intelligence. Renewable and Sustainable Energy Reviews, 77, 297–317.

Kalogirou, S. A. (2001). Artificial neural networks in renewable energy systems applications: a review. Renewable and sustainable energy reviews, 5(4), 373–401.

Kaye, L. (2011, July 1). Amazon resists pressure to participate in carbon footprint disclosure project. Retrieved from https://www.theguardian.com/sustainable-business/amazon-disclose-data-climate-carbon-footprint

Khosravi, A., Garg, S. K., & Buyya, R. (2013, August). Energy and carbon-efficient placement of virtual machines in distributed cloud data centers. In European Conference on Parallel Processing (pp. 317–328). Springer, Berlin, Heidelberg.

Kruppa, M. (2019, September 20). Palantir eyes $26bn valuation in new funding round. Retrieved from https://www.ft.com/content/0db6f058-dbef-11e9-8f9b-77216ebe1f17

Miller, R. (2019, September 20). Google is investing $3.3B to build clean data centers in Europe. Retrieved from https://techcrunch.com/2019/09/20/google-is-investing-3-3b-to-build-clean-data-centers-in-europe/

NIST. (2019, August). U.S. LEADERSHIP IN AI: A Plan for Federal Engagement in Developing Technical Standards and Related Tools.

Ongweso, E. (2019, September 18). 1,000 Protesting Uber Drivers Brought NYC Traffic to a Standstill. Retrieved from https://www.vice.com/en_us/article/59nyzk/1000-protesting-uber-drivers-brought-nyc-traffic-to-a-standstill

Pichai, S. (2019, September 20). Unleashing digital opportunities in Europe. Retrieved from https://www.blog.google/around-the-globe/google-europe/unleashing-digital-opportunities-europe/

Roston, E. (2019, June 17). Amazon, ExxonMobil Called Out Over Climate Transparency. Retrieved from https://www.bloomberg.com/news/articles/2019-06-17/amazon-exxonmobil-called-out-over-climate-transparency

Sandler, R. (2019, July 29). Amazon, Microsoft, Wayfair: Employees Stage Internal Protests Against Working With ICE. Retrieved from https://www.forbes.com/sites/rachelsandler/2019/07/19/amazon-salesforce-wayfair-employees-stage-internal-protests-for-working-with-ice/#72c8936d3e94

Statt, N. (2018, November 3). Over 20,000 Google employees participated in yesterday’s mass walkout. Retrieved from https://www.theverge.com/2018/11/2/18057716/google-walkout-20-thousand-employees-ceo-sundar-pichai-meeting

Stoller, B. (2019, February 15). Why the Northern Virginia Data Center Market Is Bigger Than Most Realize. Retrieved from https://www.datacenterknowledge.com/amazon/why-northern-virginia-data-center-market-bigger-most-realize

Strubell, E., Ganesh, A., & McCallum, A. (2019). Energy and Policy Considerations for Deep Learning in NLP. arXiv preprint arXiv:1906.02243.

Susskind, J. (2018). Future politics: Living together in a world transformed by tech. Oxford University Press.

Vidal, J. (2017, December 12). ‘Tsunami of data’ could consume one fifth of global electricity by 2025. Retrieved from https://www.climatechangenews.com/2017/12/11/tsunami-data-consume-one-fifth-global-electricity-2025/

Weaver, M., Hern, A., Bekiempis, V., Hepler, L., & Fermoso, J. (2018, November 1). Google walkout: global protests after sexual misconduct allegations. Retrieved from https://www.theguardian.com/technology/2018/nov/01/google-walkout-global-protests-employees-sexual-harassment-scandals

Alex Moltzau is a co-founder of Young Sustainable Impact and AI Social Research. Student at the University of Oslo majoring in Anthropology with a minor in Computer Science. @AlexMoltzau

Originally published at https://towardsdatascience.com/the-climate-crisis-why-data-must-not-become-the-new-oil-32d2429b8657 and republished here with permission from the author.

Photo by Markus Spiske on Unsplash