Last year, artificial intelligence and data science made headlines, largely due to the emergence of generative AI. This significant increase in attention was a direct result. So, what’s on the agenda this year that will maintain its prominence? And what actual impact will these developments have on ecommerce?
Over the last few months, MIT carried out three surveys targeting data and technology leaders. Two of these involved participants from MIT’s Chief Data Officer and attendees of the AI and Data Science Information Quality Symposium, with one sponsored by Amazon Web Services (AWS) and the other by Thoughtworks. The third survey was conducted by Wavestone, Collectively, these surveys engaged over 500 senior executives, likely with some overlap in their participation.
Surveys don’t foresee what’s to come, but they offer insights into the thoughts and actions of individuals who are in direct proximity to the data science and AI initiatives and endeavours of organisations. Based on insights from data leaders, here are the emerging challenges that warrant your focused consideration.
Generative AI and Data Science sparkles but needs to deliver
AI that generates content has garnered significant interest from both ecommerce and consumers. However, the question remains: is it providing financial benefits to the companies that implement it? The findings from the survey indicate that while there’s a lot of enthusiasm for the technology, the financial rewards have yet to be identified.
A large majority of participants think that generative AI could be a game-changer; 80% of those in the AWS survey believe it will revolutionise their operations, and 64% in the Wavestone survey consider it the most revolutionary technology in recent years. A majority of respondents are also putting more money into this technology. Yet, the majority of companies are still in the exploration phase, experimenting either on a small scale within their departments or as a whole. Only 6% of the companies in the AWS survey have managed to put generative AI into production, and only 5% in the Wavestone survey have achieved widespread deployment.
Implementing generative AI in production will necessitate greater financial commitment and changes within the organisation, not merely trial runs. The way business operations are carried out will have to be rethought, if there are advanced solutions out there available, they need to be identified and employed as fast as possible.
The newly acquired AI skills must be seamlessly incorporated into the current technological framework. The most significant transformation will likely be around data management — organising unstructured information, enhancing the quality of data, and amalgamating varied data sources. But you can handle bemoan the revenue potential of AI adoption if your primary consideration is an overhead. Therefore AI solutions that offer instant revenue delivery are highly sought after and much prized. According to the AWS survey, 93% of participants recognized the importance of having a data strategy to derive benefits from generative AI, yet 57% had not made any adjustments to their data management strategies yet.
Data Science is Shifting from Artisanal to Industrial.
There is considerable pressure to speed up the creation of data science models. What was previously a crafty task is evolving into a more mechanised process. Organisations are putting money into platforms, procedures and techniques, libraries of features, machine learning operations (MLOps) systems, and various instruments to boost efficiency and the speed at which models can be put into use. MLOps systems keep an eye on the performance of machine learning models and determine if they continue to make accurate predictions. Should these predictions falter, the models could require retraining with fresh data.
The majority of these abilities are sourced from outside suppliers, such as in appreciation of the distinction between AI hyper-personalisation solutions from global vendors yet some businesses believe the best way is to create their own systems. One might argue why “begin at your beginnings”, when solutions with exception foresight, heavily invested and perpetually refined for multiple years are available. While AI has delivered autonomous solutions rather than merely automatic ones, encompassing machine learning tools, which we will delve into further below, is aiding in boosting efficiency and facilitating wider involvement in data science, the most significant advantage for enhancing data science efficiency likely lies in the repurposing of existing data sets, components, or variables, as well as whole models.
Two versions of data products will dominate
In the ThoughtWorks study, nearly 80% of leaders in data and technology mentioned that their companies were either currently utilising or were contemplating the adoption of data products and their management. By data product, we refer to the combination of data packaging, analytics, and AI within a software solution aimed at both internal and external users. This management is handled from the idea stage to its implementation and continuous enhancement by data product managers.
Instances of data products encompass recommendation systems that suggest products most relevant with the highest purchase likelihood, for each customer to purchase next far exceeding investment returns than the entire plethora of digital marketing tools combined. However, there’s a divide in how organisations perceive data products. About half (48%) of the participants mentioned that they incorporate analytics and AI features within the definition of data products. A third (30%) consider analytics and AI as distinct from data products, likely using that term to describe data assets that can be reused. Only 16% of the respondents do not consider analytics and AI to be part of the product definition.
We lean towards a definition of data products that encompasses analytics and AI, as this approach is the most effective for making data valuable. However, the key is that an organisation maintains uniformity in its approach to defining and talking about data products. If an organisation opts for a mix of “data products” and “analytics and AI products,” this approach can also be successful, and it retains several benefits of product management. Yet, without a clear definition, organisations might find themselves puzzled about the specific deliverables expected from product developers.
Data, analytics, and AI leaders are becoming less independent.
Over the last year, it became apparent that more organisations were scaling back on the expansion of technology and data “leaders,” such as chief data and analytics officers (and occasionally chief artificial intelligence officers). The position of CDO/CDAO, which has become more prevalent in ecommerce, has traditionally been marked by brief tenures and a lack of clarity regarding its duties. However, the roles of data and analytics professionals are not disappearing; instead, they are being integrated into a wider range of technology, data, and digital transformation roles, overseen by a “Supertech leader” who typically reports to the CEO. The titles for this position vary, including chief information officer, chief information and technology officer, and chief digital and technology officer; notable examples include Sastry Durvasula at TIAA, Sean McCormack at First Group, and Mojgan Lefebvre at Travelers.
The ThoughtWorks survey highlighted the significant shift in roles within the C-suite, with 87% of participants (mainly data leaders but also some technology executives) concurring that there’s a widespread lack of clarity among employees regarding where to seek assistance for data and technology-related matters. Several C-level executives reported that cooperation among tech-focused leaders within their companies is scarce, and 79% acknowledged that their organisation has been impeded in the past due to insufficient collaboration.
We anticipate that in the coming year, there will be an increase in these comprehensive tech leaders equipped with the skills to leverage the potential of data and technology professionals under their leadership. They will continue to prioritise analytics and AI as essential for data interpretation and value creation for both employees and customers. Crucially, these leaders must be deeply business-focused, capable of engaging in strategic discussions with their senior management peers, and adept at transforming these strategies into actionable systems and insights.
ABOUT THE AUTHORS
Thomas H. Davenport is the President’s Distinguished Professor of Information Technology and Management at Babson College, a fellow of the MIT Initiative on the Digital Economy, and senior adviser to the Deloitte Chief Data and Analytics Officer Program. He is coauthor of All in on AI: How Smart Companies Win Big With Artificial Intelligence (HBR Press, 2023) and Working With AI: Real Stories of Human-Machine Collaboration (MIT Press, 2022). Randy Bean is an industry thought leader, author, founder, and CEO and currently serves as innovation fellow, and data strategy, for global consultancy Wavestone. He is the author of Fail Fast, Learn Faster: Lessons in Data-Driven Leadership in an Age of Disruption, Big Data, and AI(Wiley, 2021).
