Does Your Team Trust the Machine? The How of Securing AI Buy-In  

The Data

1. 1

   25% of earnings calls in 2024 include discussions about future plans and a vision for AI in supply chain.

2. 2

   A Zero100 survey of supply chain professionals reveals that communicating benefits, proactively addressing concerns, and side-by-side comparisons are seen as the best ways to secure team trust in AI models.

3. 3

   Trust in AI might also be achieved through the use of synthetic data – a nascent approach and technology that Amazon is currently using.

Trust the Machine 

We know AI ROI takes the form of gains like more accurate demand forecasting to waste reduction, increased efficiency to stronger trend prediction. It’s unsurprising, then, that our research found 25% of earnings calls in 2024 feature talk about the vision for AI vision supply chain. However, turning vision into a reality requires more than just the tech itself. It requires team trust and buy-in on AI. So, how might it be gained?  

Zero100 conducts an annual survey to understand global perspectives on the state of AI in supply chain. This year, one question on the survey was: “How do you secure team trust and buy-in of AI/ML models/system-generated recommendations?” The top answers were: 

• Conduct communication campaigns showing users “what’s in it for me” (63%) 

• Run stakeholder feedback sessions to proactively address concerns (62%) 

• Run old and new systems side by side to demonstrate the improved accuracy or reliability of the new system vs the legacy (51%) 

Ultimately, the how is key. Understanding that AI models produce positive results is one piece of this; knowing how it works and that it can be trusted to make decisions or recommendations is another. The results of our survey show these elements, alongside communication, are key. In reality, one way to show what AI can do, how it can help, and explaining the decisions it makes might be synthetic data.  

Synthetic Data Is Key for AI Explainability 

Synthetic data – artificially generated information that mimics real-world data – can be used to train AI models when real data is scarce, sensitive, or costly to obtain. It is created using algorithms, simulations, or models to replicate the patterns and characteristics of real data while avoiding privacy or compliance risks. Using it better allows humans to analyze data-driven decisions – rather than black box answers without much explanation of the why – made by AI.  

Amazon is one of the few companies experimenting with using synthetic data to increase explainability. Amazon SageMaker Clarify addresses two core issues in ML models: bias detection and model explainability. By identifying biases in data and predictions and offering insights into how models make decisions, Clarify ensures AI systems operate fairly and transparently. It is integrated seamlessly into existing ML workflows, allowing ongoing fairness checks from data preparation to deployment. 

The Takeaway 

Strategies to help build team trust are, broadly, all about communicating its value, which is both in terms of enhancing people’s jobs and in terms of business performance. Consider what that will look like in reality – our survey suggests communication campaigns, feedback sessions, and running sessions with and without AI side-by-side, but look consider other options, for example, involving employees in reshaping the future of work or, as we delve into above, synthetic data. There are various ways to address the specific concerns your teams may be worried about. And with all of this in mind, commit to principles of change management as this shift to a different future of work is just one part of the larger change happening in supply chain.   

Reach out to us at hello@zero100.com to learn more about our AI frameworks and the Zero100 AI Blueprint, which is how we help companies mold their data-centric supply chain strategies and prioritize AI investments. 

Methodology: 

Zero100’s proprietary data and analytics are a combined effort between our data scientists and research analysts. We provide data-first insights matched with our own research-backed points of view and bring this analysis to life via real-world case examples being led by supply chain practitioners today.   

For this study, we analyzed 2024 earnings calls for 150 brands pulling relevant insights and categorizing keywords, phrases, and sentiments. We also surveyed 312 supply chain professionals, including CSCOs and COOs. Respondents include supply chain professionals from companies with an average company revenue of over $1 billion. Seniority of position ranges from Senior Director to C-suite. Respondents work across all functions and all regions, with the majority based in the USA. 

Further Reading  

• The Signal: Making Peace with GenAI as Job Killer 

• The Zero100 Podcast: Human-Machine Teams Are the Future. But How Do They Work? 

• 100 Seconds: Tamera Fenske, Chief Supply Chain Officer at Kimberly-Clark