To what extent can Artificial Intelligence apply Physics to solve global problems?
DOI:
https://doi.org/10.29311/ndtns.vi20.4705Abstract
Generative Artificial Intelligence (GenAI) is an emerging technology that creates relevant text, images and other content from prompts. Large Language models (LLMs) are the most widely used of these GenAI forms. This technology already has applications in business and education.
This paper tests GenAI’s ability to apply physics to global problems and arrive at viable solutions. When an idea is created by a human, it is merely a culmination of that person’s experiences and prior knowledge, ordered into a new concept. This research proposes that the process should be able to be replicated by a machine learning algorithm and, due to its vast database, a far more informed and coherent idea should be the result. This research tested how well AI could tackle some global challenges and compared the results to how well these same challenges could be addressed by physicists.
The data collection process was to have a dynamic conversation with each of the participants and work with them to create a number of ideas and solutions that apply physics to a selection of global issues. This process was repeated with both Bing AI and ChatGPT-4, where they were prompted to return ideas to the same issues. Each of the ideas were then coded to a marking scheme adapted from the ‘OECD DAC criteria for development evaluation’.
While Bing AI did not prove itself to be capable of unique idea creation, ChatGPT-4 returned valuable data. ChatGPT-4 excelled at providing efficient, coherent and sustainable results whilst it performed significantly worse than humans in versatility and profitability.
The findings show that at the present time, AI cannot work as an idea generation tool on its own due to lacking in accuracy and versatility. It is best applied in tandem with humans where it can be used to generate a series of ideas to a problem and physicists refine the results.
References
Blasi, A.D. (2023). A Symphony of Senses: Introducing Sensory AI: A Pathway to Achieving Artificial General Intelligence, Medium, [Online]. Available from: https://medium.com/@aarondiblasi/a-symphony-of-senses-introducing-sensory-ai-a-pathway-to-achieving-artificial-general-8bd731154d09.
Borji, A. (2023). A Categorical Archive of ChatGPT Failures. arXiv:2302.03494 [cs].
Choi, W., Park, J., Han, D.J., Park, Y. and Moon, J. (2024). Consistency-Guided Temperature Scaling Using Style and Content Information for Out-of-Domain Calibration. arXiv.org. [Online]. [Accessed 6 March 2024]. Available from: https://arxiv.org/abs/2402.15019.
Durante, Z., Sarkar, B., Gong, R., Taori, R., Noda, Y., Tang, P., Adeli, E., Lakshmikanth, S.K., Schulman, K., Milstein, A., Terzopoulos, D., Famoti, A., Kuno, N., Llorens, A., Vo, H., Ikeuchi, K., Fei-Fei, L., Gao, J., Wake, N. and Huang, Q. (2024). An Interactive Agent Foundation Model. arXiv.org. [Online]. [Accessed 6 March 2024]. Available from: https://arxiv.org/abs/2402.05929.
He, Y., Qiu, J., Zhang, W. and Yuan, Z. (2024). Fortifying Ethical Boundaries in AI: Advanced Strategies for Enhancing Security in Large Language Models. arXiv.org. [Online]. [Accessed 6 March 2024]. Available from: https://arxiv.org/abs/2402.01725.
Hughes, O. (2023). Generative AI defined: How it works, benefits and dangers. TechRepublic. [Online]. Available from: https://www.techrepublic.com/article/what-is-generative-ai/.transforming-the-world/.
Irion, S., Christaki, U., Berthelot, H., L’Helguen, S. and Jardillier, L. (2021). Small phytoplankton contribute greatly to CO2-fixation after the diatom bloom in the Southern Ocean. The ISME Journal.
Kelly, W. (2024). GPT-3.5 vs. GPT-4: Biggest differences to consider | TechTarget. Enterprise AI. [Online]. [Accessed 6 March 2024]. Available from:
https://www.techtarget.com/searchEnterpriseAI/tip/GPT-35-vs-GPT-4-Biggest-differences-to-consider.
Kumar, M. (2023). Understanding Tokens in ChatGPT. Medium. [Online]. [Accessed 6 March 2024]. Available from:
https://medium.com/@manav.kumar87/understanding-tokens-in-chatgpt-32845987858d.
Lea, R. (2022). What is nuclear fission? Space.com. [Online]. Available from:
https://www.space.com/what-is-nuclear-fission.
Liang, Y., Ries, M.E. and Hine Peter Hine, P.J. (2022). Three methods to measure the dissolution activation energy of cellulosic fibres using time-temperature superposition, Carbohydrate Polymers, p.119541.
Mittal, A. (2023). The Essential Guide to Prompt Engineering in ChatGPT - Unite.AI. www.unite.ai. [Online]. Available from: https://www.unite.ai/prompt-engineering-in-chatgpt/.
Mittelstadt, B., Wachter, S. and Russell, C. (2023). To protect science, we must use LLMs as zero-shot translators, Nature Human Behaviour, 7(11), pp.1830–1832.
Niemela, J.J. (2021). Physics for a better world, Nature Physics, 17(8), pp.871–872.
OECD. (2023a). Evaluation Criteria-OECD. www.oecd.org. [Online].Available from:
https://www.oecd.org/dac/evaluation/daccriteriaforevaluatingdevelopmentassistance.htm.
OECD. (2023b). OECD.org - OECD. Oecd.org. [Online]. Available from: https://www.oecd.org/.
OpenAI. (2023). GPT-4 Technical Report. arXiv:2303.08774 [cs].
Redman, M., King, A., Watson, C. and King, D. (2016_. What Is CRISPR/Cas9? Archives of Disease in Childhood-Education & Practice Edition, 101(4), pp.213–215.
Renato. (2023). Enhancing Chemistry Learning with ChatGPT and Bing Chat as Agents to Think With: A Comparative Case Study. arXiv (Cornell University).
Roy, A. (2020). Introduction To Autoencoders, Medium, [Online].
Available from: https://towardsdatascience.com/introduction-to-autoencoders-7a47cf4ef14b.
Sahoo, P., Singh, A.K., Saha, S., Jain, V., Mondal, S. and Chadha, A. (2024). A Systematic Survey of Prompt Engineering in Large Language Models: Techniques and Applications, arXiv (Cornell University).
Safrai, M. and Azaria, A. (2023). Performance of ChatGPT-3.5 and GPT-4 on the United States Medical Licensing Examination with and without Distractions. arXiv.org. [Online]. [Accessed 6 March 2024]. Available from: https://arxiv.org/abs/2309.08625.
Sarma, S.D. (2023). How AI and ML Will Affect Physics, Physics, 16, p.166.
Walsh, M. 2023. ChatGPT Statistics. (2023). Essential Facts and Figures. Style Factory. [Online]. Available from: https://www.stylefactoryproductions.com/blog/chatgpt-statistics.
West, D. and Allen, J. (2018). How artificial intelligence is transforming the world. Brookings. [Online]. Available from: https://www.brookings.edu/articles/how-artificial-intelligence-is-[14]
