Machine learning (ML) permeates every corner of an organization, powering everything from internal tools that improve efficiency to customer-facing features that enrich user experiences. But once the commitment is made to launch an ML project, or any other AI projects, a new challenge emerges: effectively communicating its progress and potential impact to non-technical stakeholders.
Technical leaders — ML engineers, product managers, and project leaders — are tasked with translating intricate ML updates into digestible insights for audiences with diverse backgrounds, interests, and goals. The communication challenge requires more than technical fluency; it demands adaptability, empathy, and a strategic framework. Here, we’ll explore a robust, actionable framework to help technical teams bridge this communication gap with ease.
In today’s digital age, AI adoption is not just a trend but a necessity for business success. By leveraging AI, businesses can automate repetitive tasks, gain valuable insights from vast amounts of data, and make more informed decisions. This leads to improved operational efficiency, cost reduction, and enhanced customer experiences. For instance, AI can streamline supply chain management, optimize marketing strategies, and personalize customer interactions, giving businesses a competitive edge. Moreover, AI fosters innovation, enabling companies to develop new products and services that meet evolving market demands.
Failing to adopt AI can have significant repercussions for businesses. Without AI, companies may find it challenging to keep pace with competitors who are leveraging these technologies to their advantage. This can result in missed opportunities, such as the inability to capitalize on emerging market trends or to respond swiftly to changing customer needs. Additionally, the lack of AI adoption can lead to decreased productivity and reduced customer satisfaction, ultimately impacting the bottom line. In a rapidly evolving business landscape, staying ahead requires embracing AI to drive growth and maintain relevance.
Machine learning is a subset of artificial intelligence that focuses on training algorithms to learn from data and make predictions or decisions without explicit programming. It’s the driving force behind many AI applications, from image and speech recognition to natural language processing and predictive analytics. By analyzing large datasets, machine learning models can identify patterns and provide actionable insights, helping businesses make data-driven decisions.
Machine learning models are trained using various algorithms, such as support vector machines, decision trees, and neural networks. These models can perform tasks like classification, regression, clustering, and dimensionality reduction. Deep learning, a type of machine learning, involves neural networks with multiple layers and is particularly effective for tasks involving image and speech recognition. Reinforcement learning, another branch of machine learning, trains agents to make decisions in complex environments.
The journey to smooth communication starts with designating a single point of contact (PoC). By integrating data science, businesses can further enhance their AI capabilities and gain deeper insights from their data. This person becomes the bridge between the technical team and stakeholders, overseeing how updates are presented, organized, and translated for each audience. Selecting the PoC often falls to a product manager due to their unique balance of technical insight and communication skills.
But what’s crucial here isn’t just a skill for translating tech jargon but for curating relevant information tailored to the interests and investment areas of each stakeholder. Your PoC should meet with ML engineers to understand ongoing technical progress and challenges and synthesize this information to align with business metrics.
Non-technical stakeholders can encompass anyone from finance to HR and legal, each with varying concerns. Finance might grasp data pipelines and regression models, while HR may be focused on user impact, and legal on data security. Context matters immensely.
For finance, it could mean discussing project scalability and efficiency improvements; for legal, addressing data integrity or privacy compliance. Machine Learning Engineer Vitalijus Cernej, of NineTwoThree, highlights this balance:
“It’s crucial to find the balance between saying a lot of technical words and speaking in common terms.”
Computer vision techniques are used for tasks like image categorization, face detection, and object identification.
This strategy centers on anticipation. First, consider what the audience already understands, the questions they’re likely to have, and the type of medium they prefer for updates. Deep learning models, a type of machine learning, involve neural networks with multiple layers and are particularly effective for tasks involving image and speech recognition. Whether through Slack, emails, or live presentations, using a familiar medium will keep them engaged and receptive. Reinforcement learning models, another branch of machine learning, train agents to make decisions in complex environments.
Machine learning projects evolve in phases, and the format of updates should adapt accordingly. At the outset, updates might be conceptual, explaining the goal of the machine learning project and any setup completed. At the mid-stage, showing a live demo or even a mock-up of the model’s early functionality is ideal.
As NineTwoThree’s MLOps Engineer, Jurgis Samaitis, advises, “Updates can vary. Initially, it can be as simple as letting everyone know the environment is set up. Then, when you have a working demo, it might shift to demonstrating the solution answering a basic question.” Structuring updates to match project progress helps stakeholders see both the big picture and specific achievements, allowing them to ask relevant questions and offer input.
Consider this outline for various ML project stages:
One of the most effective ways to keep non-technical stakeholders involved is to engage them in conversation rather than just present information. Discussing the specific machine learning algorithms being used can help stakeholders understand the technical challenges and decisions involved. Here, the PoC can ask for feedback on project direction, solicit ideas for optimizing impact, or present multiple options for project pathways.
Openly discussing choices — whether around resource allocation, model refinement, or integration with existing tools — brings the conversation to a strategic level, where stakeholder input adds business value. This shift keeps the focus on the project’s real-world impact instead of technical details alone. Key questions to frame the discussion include:
As the ML project matures, regular internal meetings play a pivotal role in aligning expectations and gathering feedback. Machine learning professionals play a crucial role in these meetings, providing technical insights and ensuring that updates are accurate and relevant. Weekly or bi-weekly calls help the PoC keep all project members updated and refine the communication style for each new phase.
During these calls, the PoC can prepare a checklist that includes past updates, current progress, and upcoming tasks, linking each item to its impact on the project’s KPIs. Sharing a simple template can help guide these conversations, aligning all team members and stakeholders.
According to Vitalijus, “At the beginning, you’ll be speaking in a lot of abstracts… At the end, you’ll hopefully have working demos, presentations, and charts. Those are all extremely helpful.” By aligning update depth with project phase, stakeholders understand the scope better and gain a realistic view of progress.
In machine learning, setting the right expectations from the outset is essential. Advanced machine learning projects often require iterative tuning, data cleaning, and testing to approach optimal performance levels. Stakeholders might assume that ML models reach flawless accuracy upon launch. However, ML projects often require iterative tuning, data cleaning, and testing to approach optimal performance levels. Jurgis cautions that aiming for overly high accuracy too soon is not realistic.
“Some companies like to brag about their top-of-the-line accuracy,” he notes, “but that only happens after years of fine-tuning models.”
Explaining the iterative nature of ML development, and preparing stakeholders for potential hurdles, can help set reasonable expectations. Emphasize that statistics can be misleading when taken out of context, so stakeholders understand that a 70% accuracy rate can still signal success, depending on project goals and dataset quality.
Transparency around expectations minimizes potential friction and aligns stakeholders with the realities of machine learning.
