Innovation & AI – Let’s Use These Words Only When They Truly Apply
- mentallurgical
- Jul 14
- 3 min read
Updated: Aug 2
In clinical research, especially within CROs, two terms are used a lot these days- innovation and artificial intelligence (AI).
These words carry weight and create a feeling of leap. But lately, they are being thrown around indiscriminately, often used to describe routine tasks or basic automation. When everything is labeled as "innovative" or "AI-powered," it becomes difficult to tell what's genuinely impactful.
Innovation as Real Change and Not Just More Ideas
A team member recently told me that their manager asked everyone to submit three innovation ideas per week. That really made me pause. To my understanding, innovation isn’t about volume and definitely not a checklist. It’s about solving real problems in new and meaningful ways.
Yet in our day-to-day work, the term gets used for things as simple as reformatting a tracker or creating a new naming convention. These can be good process improvements, but to be honest, they are not innovations.
What’s missing in most of these examples is impact. Real innovation leads to measurable change like time saved, errors reduced etc. Not every tweak qualifies as an innovation. A BOT, originally introduced to freeze data efficiently but fails to do so which impacts delivery can at best be called "unsuccessful" innovation.
AI: Not Every Script is Intelligent
The misuse of AI is even more widespread. In clinical data science, basic automation or scripting is often labeled as AI. For example: a SAS or Python script that refreshes reports each month; a scheduled job that emails listings; rule-based logic that assigns queries to reviewers. These are automations, not AI.
Staying true to it's definition, AI systems learn, adapt, or make intelligent decisions through methods like machine learning or natural language processing. If it’s just following a fixed rule or sequence, it’s not AI.
Cases Where Terms Like 'AI' and 'Innovation' Are Used Loosely
You don’t have to look far to find examples of this confusion in industry blogs and posts:
eClinical Solutions - https://www.eclinicalsol.com/blog/ai-ml-and-me-oh-my-how-ai-ml-can-elevate-the-clinical-data-managers-role/
This blog by eClinical Solutions discusses how AI and machine learning can elevate the role of clinical data managers. While the post introduces AI/ML terminology, most examples focus on automation, rule-based workflows, and process efficiency rather than true learning systems. This highlights a common industry trend where automation is sometimes labeled as AI, even when no adaptive intelligence is involved.
A LinkedIn article https://www.linkedin.com/pulse/role-ai-clinical-data-management-enhancing-query-smart-jangili-i65pe talks about how AI could improve future query management by learning from historical data. The vision aligns with AI principles, though the post doesn't go into technical depth on how the AI is built or deployed. While the theory is sound, the lack of details on actual models or deployment makes it hard to judge if this is applied AI or just aspirational.
A PharmaSUG 2025 paper https://pharmasug.org/proceedings/2025/MM/PharmaSUG-2025-MM-085.pdf thoughtfully explores how generative AI tools can assist programmers in drafting SAS code for listings and TLFs. The authors clearly present GenAI as a support tool that still requires human review and validation. The paper is responsible and transparent in its scope. However, in broader discussions across the industry, such GenAI use cases are sometimes described as AI-driven programming, which may lead to the impression of end-to-end automation or learning systems—when in reality, the AI is providing one-time output and not adapting or learning from the data environment.
These tools are valuable, but they should be called what they are, be it automation or scripting or process optimization and perhaps not AI.
Why This Matters and Is Important
Using the right terminology isn’t just semantics, it shapes expectations and trust. Calling something AI when it isn’t fully in all sense, may mislead stakeholders. Labeling small changes as innovation can reduce credibility. Further, using the correct terms helps recognize true breakthroughs when they happen.
Innovation and AI are powerful words and they should remain that way. Let us not dilute their meaning by using them loosely or where they don’t apply. Automating a listing refresh is great, but it’s not AI. Reorganizing a spreadsheet is helpful, but it’s not innovation. We don’t need to oversell the work that we do. Honest labels, clear communication, and realistic expectations go a long way, especially in our scientific industry, where clarity matters because we are dealing with people and their lives.
Share your thoughts or examples in the comments. Have you seen other buzzwords misused in our field?