Why Citizen Data Scientists Should Not Work on AI Projects
- By Paul Mah
- December 14, 2022
The conventional wisdom is that only highly trained data scientists and AI engineers could effectively drive the development of AI strategy and implementation. However, this belief is increasingly being challenged by the development of auto-ML software and other advanced tools designed to facilitate the implementation of AI.
This has led to the idea of engaging citizen data scientists in the form of domain experts and other team members highly familiar with the business processes to accelerate AI efforts.
But while opening the floodgates of AI can bring many benefits, a new report in the Harvard Business Review cautioned that there are also potential risks associated with making data science and AI more accessible to non-technical professionals.
The risks of democratizing AI
There are primarily three risks, with the first being the inability of auto-ML tools to solve gaps in expertise, training, and experience. Specifically, auto-ML was meant as a tool for data scientists to write code quickly – a non-expert using it would not be capable of spotting pitfalls and situations where the resulting AI can break.
One example cited was the inability of a citizen data scientist to handle unbalanced training data sets. In many instances, dubious transactions of just 1% could render a data set useless for the AI project. However, an employee not trained in data science would be unable to properly tailor the data sample or end up testing the AI model against the wrong benchmark.
In addition, the field of AI is also rife with ethical, reputational, regulatory, and legal risks. This is an area that even experienced data scientists and AI experts are hard-pressed to address, much less AI novices.
Indeed, a study conducted by Oxford Economics for the IBM Institute for Business Value revealed a radical shift in the roles responsible for leading and upholding AI ethics at an organization. And underscoring how trustworthy AI is now perceived as a strategic differentiator, the majority (80%) of respondents pointed to a non-technical executive as the primary “champion” for AI ethics, a sharp uptick from 15% in 2018.
Finally, a citizen data scientist might have less awareness when a particular AI initiative will not work: “[Having] AI novices spend time developing AI can lead to wasted efforts and internal resources on projects better left on the cutting room floor. And potentially worse than that, faulty models that get used may lead to significant unforeseen negative impacts.”
Paul Mah is the editor of DSAITrends. A former system administrator, programmer, and IT lecturer, he enjoys writing both code and prose. You can reach him at [email protected].
Image credit: iStockphoto/PrathanChorruangsak
Paul Mah
Paul Mah is the editor of DSAITrends, where he report on the latest developments in data science and AI. A former system administrator, programmer, and IT lecturer, he enjoys writing both code and prose.