Pair Programming in AI Data Analysis: Benefits and Limitations
How can we benefit from pair programming? Could it become an educational process for inexperienced team members?
By Fannie Thoo
Microsoft and AWS have detailed papers on what a data product is and how data can be treated as a product (I have shared links to both at the bottom). According to both, data product is commonly associated with the data mesh architecture pattern, where it embraces the idea of treating data as a product. By continuously improving and enhancing data product structures, access, and policies, organizations can unlock the full potential of their data through analytics.
Given the importance of programming abilities in delivering data products, the adoption of pair programming can greatly benefit freshies in data teams. Through this collaborative approach, inexperienced team members can expedite hidden insight discovery journey, ultimately leading to advantages for the data teams.
AI-assisted pair programming is a great way for beginners to improve their coding skills, says Rand-Hendriksen in his training “Pair Programming with AI Online Class”. Several other sources agree that Generative AI tools enable fresh data team members with limited technical experience to learn and benefit from the expertise of their more experienced teammates.
Application of these effective educational techniques along with implementation of agile software development will enhance the problem-solving prowess of data teams throughout their research and data product delivery expedition. To illustrate, the combination of Geospatial Artificial Intelligence (GeoAI) and GenAI has the potential to expedite the assimilation of interdisciplinary expertise and proficiencies across diverse fields.
For instance, data teams can effectively execute geospatial time series analysis by utilizing the Backseat Driver pair programming approach. This approach helps to reduce the duration of data analytics that require knowledge from various disciplines including geography, GI Science, computer science, data science, remote sensing, Earth system science, urban planning, civil engineering, and public health, according to the article “A five-year milestone: reflections on advances and limitations in GeoAI research” by Hu, Yingjie et al. Through effective implementation of this pair programming approach, the freshies in data teams can leverage the advice, guidance or feedback from their seniors to expedite the delivery of data product.
While acknowledging the vast possibilities of this combination, it is crucial to devise strategies aimed at mitigating the drawbacks of AI, particularly the occurrence of hallucinations and bias in selected AI tools.
In summary, by leveraging both human touch and cognitive abilities, we can fully harness the immense potential of AI, so long their known limitations are well and proactively managed. The more aware we are about these limitations, the better we will benefit from AI, utilizing tools that can manage these limitations with personalized approach.
References:
AWS. (2023). Unlock the Power of Data as a Product How an end-to-end data strategy supports productization. In AWS. Amazon Web Services, Inc. or its Affiliates. https://pages.awscloud.com/rs/112-TZM-766/images/PTNR-AWS-DataProduct-eBook.pdf
Bavya, B. K. T., & Infosys Limited. (2024). Benefits of Pair Programming with Generative AI. Blogs.infosys.com; Infosys. https://blogs.infosys.com/digital-experience/emerging-technologies/pair-programming-and-benefits.html
Code, C. (2023, June 15). The Ultimate Guide to Effective Pair Programming. DEV Community; Substack. https://dev.to/evergrowingdev/the-ultimate-guide-to-effective-pair-programming-5aej
Codeacademy Team. (2021, September 24). What Is Pair Programming? Codecademy. https://www.codecademy.com/resources/blog/what-is-pair-programming/
Hu ,Yingjie, Michael, G., Zhu ,AXing, May, Y., Aydin ,Orhun, Bhaduri ,Budhendra, Song, G., Wenwen, L., Dalton, L., & Shawn, N. (2024). A five-year milestone: reflections on advances and limitations in GeoAI research. Annals of GIS, 30(1), 1–14. https://doi.org/10.1080/19475683.2024.2309866
Microsoft. (2022, April 19). What is a data product? — Cloud Adoption Framework. AI Skills Challenge; Microsoft. https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/scenarios/cloud-scale-analytics/architectures/what-is-data-product
MIT Sloan Teaching & Learning Technologies. (2024). When AI Gets It Wrong: Addressing AI Hallucinations and Bias. MIT Sloan Teaching & Learning Technologies. https://mitsloanedtech.mit.edu/ai/basics/addressing-ai-hallucinations-and-bias/
Rand-Hendriksen, M. (2023, April 3). Pair Programming with AI Online Class | LinkedIn Learning, formerly Lynda.com. LinkedIn. https://www.linkedin.com/learning/pair-programming-with-ai
