(US & Canada) | Using AI Without Understanding Data Can Lead to Precarious Situations — The Boldt Company CIO and CTO

Maribeth Achterberg, CIO and CTO at The Boldt Company, speaks with Thiago Da Costa, CEO of Toric, in a video interview, about organizational strategies to foster data and analytics culture, the importance of having the right metrics, leveraging data while ensuring data protection, the role of regulations, understanding data prior to doing AI, investing in the existing technologies before leveraging GenAI, and having AI use policy.

The Boldt Company is a 134-year-old construction management company based in Appleton, Wisconsin, with 17 offices across the U.S.

Speaking about how leadership can foster data and analytic culture, Achterberg says that the management must measure the things that matter. Speaking of the most crucial metrics in the construction space, she mentions exit margin erosion, apart from revenue.

Achterberg argues that all metrics have the same concept, and it requires understanding the contributing processes, supporting information, and information systems, and bringing all of them together to calculate the metric.

Improving the metric improves the profit margin from a company standpoint, says Achterberg. Therefore, she urges executives in construction and other industries to pivot people’s focus to metrics that matter and educate them on how they contribute to it.

To get to that metric, it is imperative to have a solid handle on organizational data in aggregate, says Achterberg. She reiterates the need to lay out critical metrics and build a strategy to use data against those metrics. It all boils down to value creation and measuring what matters to get the outcome and have an evolved culture.

Taking the safety aspect in construction as an instance, Achterberg discusses the technology used around safety metrics to ensure minimal accidents. This could include putting sensors on a vest or on hard hats to track people on-site to make sure they are not somewhere they should not be.

Further, Achterberg mentions exoskeletons that ensure proper posture and reduce stress wear-and-tear on the human body. She maintains that while many such technologies can be adopted but would not serve any organizational purpose unless the critical metrics are defined.

Next, Achterberg comments on striking a balance between leveraging data for decision-making and ensuring data protection and compliance. In construction management, she says, many outside entities are involved, and the organization has to manage all their personal information and certifications.

Additionally, as many states in the U.S. are adopting protective legislation such as PII, managing a person’s personally identifiable information becomes intricate, notes Achterberg. She conveys her concern about data protection when it comes to HRIS systems — where the information should reside and who should access it.

Achterberg stresses that it boils down to having the right expertise to organize the organization better, like in a hub-and-spoke operating and organizational model. For instance, regulations specific to HR must sit in HR with union and labor relations.

Consequently, the regulations push better integration across functional silos because the regulations are becoming more data-specific, showing how data can be leveraged to protect individuals.

In continuation, Achterberg states that regulations protect individuals, and being a business that leverages people’s information, protecting that information must be a priority.

Moving forward, Achterberg states that it is crucial to figure out the basics before going for the next shiny thing. Referring to GenAI and LLMs, she says that those will operate in the private data environment, therefore, the organization must understand its data first.

Furthermore, Achterberg insists on getting a grasp on governance, at least to understand where the data comes from. Otherwise, it would be a challenge to know if the LLMs hallucinate or provide the wrong feedback.

While AI or GenAI is no magic bullet, it does help organizations leapfrog over old technology and think past old data warehousing architecture and approaches, says Achterberg. It allows flexibility in solving business problems, unlike in the past. Nevertheless, using AI without understanding data will lead organizations to precarious situations, she asserts.

From a data architecture standpoint, Achterberg encourages organizations to leapfrog the old and start with the new. She states that organizations should approach it by choosing a business problem that can be solved and having a solid definition of the problem.

Instead of running to LLMs, Achterberg recommends considering digital workers and assessing how employing robotic process automation can generate organizational benefits. She stresses evolving simple processes to reduce human factors and use a bot, allowing humans to do a higher level of work.

In addition, Achterberg suggests that business efficiencies can be created in many ways by leveraging technologies that have been in the market before heading straight to GenAI.

From the data privacy perspective, Achterberg cautions organizations against putting proprietary company information into the LLMs. She confirms publishing a “Generative AI Acceptable Use” policy for the organization to put this into effect and states that all the tools allowed in Boldt asset require her approval.

In conclusion, Achterberg opines that one has to be careful about what is being exposed in the public large language models. She recalls speaking at an AI summit after the release of ChatGPT, wherein, she focused on the ethics around generative AI and what should be the society’s concern.

CDO Magazine appreciates Maribeth Achterberg for sharing her insights with our global community.