“Just Do Something with AI”: Bridging the Business Communication Gap for ML

As a machine learning (ML) practitioner, you’ve probably encountered the inevitable request: “Can we do something with AI?” This question, often posed by executives or business stakeholders, reflects both excitement and uncertainty about artificial intelligence’s potential.

Stephanie Kirmer, Senior Machine Learning Engineer at DataGrail, addresses this challenge in her talk, “Just Do Something with AI: Bridging the Business Communication Gap for ML Practitioners.” The key takeaway? Successfully integrating AI into business operations isn’t just about technical expertise — it’s about clear communication, strategic decision-making, and understanding business needs.

This blog explores how ML practitioners can navigate these conversations, ensuring AI initiatives align with real business value.

Avoiding Common Pitfalls: How Not to Respond

When confronted with a vague AI request, many practitioners instinctively react in ways that, while understandable, can be counterproductive:

• Pivoting immediately to the new AI idea, abandoning current work.
• Ignoring the request, hoping it fades away.
• Rejecting the idea outright without investigation.

Each of these responses can damage credibility. Constantly shifting focus disrupts productivity, while outright rejection can make you seem dismissive. Ignoring the request risks being perceived as disengaged. Instead, a more strategic approach is needed.

The Right Approach: Asking the Right Questions

Rather than reacting impulsively, take a step back and ask key questions to assess the proposal’s validity:

• Who is asking? Is this request coming from a decision-maker, a marketing team member, or a tech-savvy colleague? Understanding the source helps tailor your response.
• What is their motivation? Are they responding to external pressures, internal trends, or a specific pain point?
• Why now? Is there an urgent problem AI is expected to solve, or is this just an exploratory initiative?
• What connections are they seeing? Are they referencing competitors, industry trends, or personal experience?

These questions help uncover whether the request stems from a genuine business problem or is simply a reaction to AI hype.

Defining the Problem and Business Value

AI should never be implemented just for the sake of using AI. Instead, practitioners should define the business problem first and assess whether AI is the best tool for the job.

For example, suppose an executive wants to extract data from handwritten forms using AI. Before diving into model development, consider:

• What specific value will this solution provide?
• Are there existing manual or rule-based approaches that are sufficient?
• How will AI integrate into current workflows?

AI is not an inherently good solution — it’s simply a tool. Without a well-defined problem, efforts risk being wasted on solutions that offer little tangible benefit.

AI vs. Simpler Alternatives: Choosing the Right Solution

One of the most crucial steps is determining whether AI is necessary. Many business problems can be solved more efficiently with simpler automation techniques.

For instance, if-then rule-based systems or basic scripting might address the issue without the complexity, cost, and risks of AI. Additionally, business stakeholders often have a loose definition of “AI” and may not distinguish between simple automation, classical ML models, and deep learning techniques. Clarifying expectations ensures alignment before significant resources are committed.

Risk vs. Reward: Understanding Potential Downsides

Every AI implementation carries risks — technical, ethical, and legal. Before proceeding, ask:

• Could the AI system introduce errors? For example, an OCR model for handwritten text may misinterpret words, leading to costly mistakes.
• Are there privacy concerns? AI applications involving personal data must comply with legal regulations like GDPR.
• Is generative AI involved? If so, unintended or biased outputs could pose reputational risks.

A responsible approach considers these risks before moving forward.

The Decision Point: Is AI the Right Move?

Before proceeding, ML practitioners should assess:

• Is the problem well-defined?
• Is solving it a priority?
• Does everyone understand what AI can (and can’t) do in this context?
• Are the risks manageable?

If the answer to any of these is uncertain, it may be wise to pause or pivot. Not every problem requires an AI-driven solution.

Planning, Scope, and Opportunity Cost

If AI is the right approach, the next step is strategic planning. This includes:

• Identifying the required resources, skills, and technology.
• Developing a roadmap for a Minimum Viable Product (MVP) or Proof of Concept.
• Assessing trade-offs — what existing priorities will be delayed by investing in AI?

Stakeholders should understand that AI projects often require long-term commitment and iteration rather than immediate results.

Final Decision and Clear Communication

Once a decision is made — whether to move forward, pivot to another solution, or shelve the idea — it must be clearly communicated. Transparency ensures stakeholders understand the rationale behind the choice.

If the project moves forward, set expectations about timelines, challenges, and outcomes. If it doesn’t, explain why it wasn’t the right fit at this time, reinforcing that the decision was thoughtful rather than dismissive.

The Benefits of a Strategic Approach

Handling AI requests strategically improves both team reputation and business outcomes. It positions ML practitioners as trusted advisors, rather than mere executors of vague AI initiatives.

By focusing on well-defined problems, aligning AI with business value, and considering risks and resources, teams can ensure that AI initiatives are meaningful, impactful, and sustainable.

Conclusion

The growing demand for AI & AI business communication means ML practitioners must not only be technical experts but also skilled communicators. Successfully bridging the AI-business gap requires asking the right questions, defining value, managing risk, and making informed decisions.

If you want to dive deeper into AI strategy, business alignment, and the latest in machine learning, join us at ODSC East in Boston! Connect with industry leaders, gain hands-on experience, and explore cutting-edge AI solutions that drive real business impact.

ODSC East | Boston, April 23–25, 2024 |

Learn more and register at odsc.com/boston