Abstract
Humanity is currently facing one of its biggest challenges to date: The climate crisis. As a result, most industry sectors are reassessing their ways of working to be better equipped to address their share of the situation. The digital sector often gets set aside in such considerations in talk about the green transition because a significant amount of its work consists of optimizing processes that can save resources. Deep neural networks (DNNs) have gained great attraction and have shown good results regarding process automation. We argue that there are well-known and lesser known negative side-effects to automation frameworks based on DNNs (and related technologies) in terms of energy consumption, pollution, and social equality, that must be questioned. We analyze the operating principles and deployment methods of DNNs, the new era of automation efforts this has launched, and argue on this basis that their continued use is both unsustainable and indefensible. Using three examples of ongoing research, we explain how alternative approaches to develop more general machine intelligence are well-poised to power the next phase of AI-based automation.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
According to Statista, IT-related revenue has a predicted annual growth rate of 6.86% and a predicted market volume of US$1,570.00bn by 2027 (https://www.statista.com/outlook/tmo/it-services/worldwide – accessed March 1st, 2023).
- 2.
Our use of the term ‘contemporary AI’ refers to a set of methodologies that are currently in active experimentation or use in industry, including but not limited to reinforcement learning, ANNs of all kinds, and other well-known methods.
- 3.
For instance, the annual prediction that “full self-driving cars will be available next year” has been updated at a rate of one year per year by Tesla’s CEO (“Watch Elon Musk Promise Self-Driving Cars ‘Next Year’ Every Year Since 2014,” https://futurism.com/video-elon-musk-promising-self-driving-cars — accessed March 1st, 2023.
- 4.
The UN defines ‘sustainable development’ as harmony between economic growth, social inclusion, and environmental protection. https://www.un.org/sustainabledevelopment/development-agenda/ — accessed April 4th, 2023.
- 5.
ChatGPT has passed the Wharton Exam, US medical licensing exam, law school exam, and others. (https://www.businessinsider.com/list-here-are-the-exams-chatgpt-has-passed-so-far-2023-1?r=US&IR=T#wharton-mba-exam-1 — accessed March 4th, 2023).
- 6.
As of April 2023, the price is $20 a month for reliable and fast access to ChatGPT, although a free version with slower response is still available. (https://openai.com/blog/chatgpt-plus — accessed April 4th, 2023).
- 7.
By ‘knowledge’ we mean a form of ‘actionable information’—that is, information that can be used for making plans and getting things done in a particular environment.
- 8.
The AERA system, for example, learned to do a TV-style interview after learning for only 20 h on a 6-core office desktop machine [35].
References
What is the carbon footprint of a laptop? Circular Computing (2021). https://circularcomputing.com/news/carbon-footprint-laptop/
Altman, S.: Moore’s law for everything (2021). https://moores.samaltman.com/
Ananthaswamy, A.: In AI, is bigger always better? Nature 615, 202–205 (2023)
Bender, E.M., McMillan-Major, A., Gebru, T., Shmitchell, S.: On the dangers of stochastic parrots: can language models be too big?. In: Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, pp. 610–623 (2021)
Brown, T.B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., et al.: Language models are few-shot learners. Adv. Neural Inf. Process. Syst. 33, 1–25 (2020)
Butler, S.: From retail to transport: how ai is changing every corner of the economy. The Guardian (2023). https://www.theguardian.com/technology/2023/feb/18/from-retail-to-transport-how-ai-is-changing-every-corner-of-the-economy
Eberding, L.M., Thórisson, K.R., Prabu, A., Jaroria, S., Sheikhlar, A.: Comparison of machine learners on an aba experiment format of the cart-pole task. In: Proceedings of Machine Learning Research, vol. 159, pp. 49–63 (2021)
Hammer, P.: Reasoning-learning systems based on non-axiomatic reasoning system theory, vol. 192, pp. 88–106 (2022)
Hunt, E.: Tay, microsoft’s AI chatbot, gets a crash course in racism from twitter. The Guardian (2016). https://www.theguardian.com/technology/2016/mar/24/tay-microsofts-ai-chatbot-gets-a-crash-course-in-racism-from-twitter
Kaplan, J., et al.: Scaling laws for neural language models (2020)
Klinger, J., Mateos-Garcia, J.C., Stathoulopoulos, K.: A narrowing of AI research? SSRN Electron. J. (2020)
Kommrusch, S., Minsky, H., Minsky, M., Shaoul, C.: Self-supervised learning for multi-goal grid world: Comparing leela and deep q network. In: Proceedings of Machine Learning Research, vol. 131, pp. 81–97 (2020)
Latapie, H., Gabriel, M., Kompella, R.: Hybrid AI for IoT actionable insights & real-time data-driven networks. In: Proceedings of Machine Learning Research, vol. 159, pp. 127–131 (2022)
Lenat, D.: Cyc: a large-scale investment in knowledge infrastructure. Commun. ACM 38, 33–38 (2023)
Logunova, I.: A guide to one-shot learning (2022). https://serokell.io/blog/nn-and-one-shot-learning
Luccioni, A.S., Viguier, S., Ligozat, A.L.: Estimating the carbon footprint of bloom, a 176b parameter language model (2022). https://arxiv.org/abs/2211.02001
Marcus, G.: Deep learning is hitting a wall. Nautilus (2023). https://nautil.us/deep-learning-is-hitting-a-wall-238440/
Marcus, G., Davis, E.: How not to test gpt-3 (2023). https://garymarcus.substack.com/p/how-not-to-test-gpt-3
Marr, B.: 13 mind-blowing things artificial intelligence can already do today. Forbes (2019). https://www.forbes.com/sites/bernardmarr/2019/11/11/13-mind-blowing-things-artificial-intelligence-can-already-do-today/?sh=39e42add6502
Maslej, N., Fattorini, L., Brynjolfsson, E., et al.: Artificial Intelligence Index Report 2023. Technical Report, Stanford University (2023). https://aiindex.stanford.edu/report/
Mok, A.: AI is expensive. a search on Google’s chatbot Bard costs the company 10 times more than a regular one, which could amount to several billion dollars. Business Insider (2023). https://www.businessinsider.com/ai-expensive-google-chatbot-bard-may-cost-company-billions-dollars-2023-2
Nielsen, K.H.: Vindmøllens historie: Sådan tæmmede danskerne vindens energi. ForskerZonen (2018). https://videnskab.dk/forskerzonen/teknologi-innovation/vindmoellenshistorie-saadan-taemmede-danskerne-vindens-energi
Nivel, E., Thórisson, K.R., et al.: Autonomous acquisition of natural language. In: IADIS International Conference on Intelligent Systems & Agents, pp. 58–66 (2014)
Nivel, E., Thórisson, K.R., Steunebrink, B., Dindo, H., et al., G.P.: Autocatalytic endogenous reflective architecture. Tech report RUTR-SCS13002, Reykjavik University - School of Computer Science (2013)
Panch, T., Mattie, H., Celi, L.A.: The ‘inconvenient truth’ about AI in healthcare. NPJ Dig. Med. 2, 77 (2019)
Pollock, J.L.: Defeasible reasoning and degrees of justification. Argu. Comput. 1(1), 7–22 (2010)
Ritchie, H., Roser, M., Rosado, P.: Energy. Our World in Data (2022). https://ourworldindata.org/energy-mix
Roose, K.: A conversation with Bing’s chatbot left me deeply unsettled. The New York Times (2023). https://www.nytimes.com/2023/02/16/technology/bing-chatbot-microsoft-chatgpt.html
Sheikhlar, A., Thórisson, K.R., Eberding, L.M.: Autonomous cumulative transfer learning. In: International Conference on Artificial General Intelligence, pp. 306–316 (2020)
Sheikhlar, A., Thórisson, K.R., Thompson, J.: Explicit analogy for autonomous transversal learning, vol. 192, pp. 48–62 (2022)
Steunebrink, B., Swan, J., Nivel, E.: The Holon system: artificial general intelligence as ‘work on command’. In: Proceedings of Machine Learning Research, vol. 192, pp. 120–126 (2022)
Stojnic, G., Gandhi, K., Yasuda, S., Lake, B.M., Dillon, M.R.: Commonsense psychology in human infants and machines. Cognition 235, 105406 (2023)
Strubell, E., Ganesh, A., McCallum, A.: Energy and policy considerations for deep learning in NLP. In: 57th Annual Meeting of the Association for Computational Linguistics (ACL) (2019)
Thórisson, K.R.: A new constructivist AI: from manual construction to self-constructive systems. In: Wang, P., Goertzel, B. (eds.) Theoretical Foundations of Artificial General Intelligence, vol. 4, pp. 145–171 (2012)
Thórisson, K.R.: Seed-programmed autonomous general learning. In: Proceedings of Machine Learning Research, vol. 131, pp. 32–70 (2020)
Thórisson, K.R., Bieger, J., Li, X., Wang, P.: Cumulative learning. In: Hammer, P., Agrawal, P., Goertzel, B., Iklé, M. (eds.) AGI 2019. LNCS (LNAI), vol. 11654, pp. 198–208. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-27005-6_20
Thórisson, K.R., Minsky, H.: The future of AI research: ten defeasible ‘axioms of intelligence’. In: Proceedings of Machine Learning Research, vol. 192, pp. 5–21 (2022)
Thórisson, K.R.: The ‘Explanation Hypothesis’ in general self-supervised Learning. In: Proceedings of Machine Learning Research, IWSSL-21, pp. 5–27 (2021)
Wang, P.: Rigid Flexibility: The Logic of Intelligence, vol. 34. Springer, Heidelberg (2006). https://doi.org/10.1007/1-4020-5045-3
Wei, J., et al.: Chain-of-thought prompting elicits reasoning in large language models (2023)
Acknowledgment
This work was supported in part by Cisco Systems, the Icelandic Institute for Intelligent Machines and Reykjavik University.
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Vallentin, A., Thórisson, K.R., Latapie, H. (2023). Addressing the Unsustainability of Deep Neural Networks with Next-Gen AI. In: Hammer, P., Alirezaie, M., Strannegård, C. (eds) Artificial General Intelligence. AGI 2023. Lecture Notes in Computer Science(), vol 13921. Springer, Cham. https://doi.org/10.1007/978-3-031-33469-6_30
Download citation
DOI: https://doi.org/10.1007/978-3-031-33469-6_30
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-33468-9
Online ISBN: 978-3-031-33469-6
eBook Packages: Computer ScienceComputer Science (R0)