AI red flags that alert healthcare supply chain to warning signs
Fielding questions to ask about AI adoption, implementation, insurgence
Last of two parts
By Rick Dana Barlow
To many, artificial intelligence (AI) either represents the answer to or cause of all of life's problems. Healthcare supply chain executives, leaders and professionals remain intimately familiar with the drill as they've experienced it before where a newly emerging technology empties the room of oxygen based on its promise and promotion. Let us count some of the ways: Robotics (including automated guided vehicles (AGV) and automated mobile robots (AMR)), robotic process automation (RPA), internet-based online exchanges, internet of things (IoT), machinelearning and machine-to-machine (M2M) communications and interoperability, drones, blockchain, radiofrequency identification chips and tags, 3-D printing, cloud computing, augmented reality (AR), virtual reality (VR), wearable and wireless computing and sensors, digital flatscreens, interactive kiosks and motion- or voice-activated signage.
And this represents the first quarter of the 21st century alone, give or take a few years before Y2K.
Exciting stuff, for sure. But as the famous motto of Marvel Comics' the Amazing Spider-Man goes, "with great power comes great responsibility." Just replace power with technology.
During the last few months, healthcare and other industries have seen something of a backlash against AI in a fever pitch as proponents and opponents of implementing "guardrails" to control the technology alternately tap the breaks and slam down on the throttle for development.
While AI seems to generate ample promise, the technology also erupts with red flags and warning signs that give some pause but encourage others — particularly those who may be heavily invested in the technology's future or who turn a risk-averse blind eye to potholes, roadblocks and speedbumps that may threaten business, culture, finance, healthcare, life and society.
Healthcare supply chain executives, leaders and professionals who are part of the Bellwether Community, by and large, can navigate through the billowing clouds of smoke surrounding AI and wafting through the industry and profession. For them, the opportunities seem promising, but the blinders wisely have been discarded. For them, AI can be a useful tool more than a cultural tormentor.
Lofty, unrealistic expectations
When it comes to AI, red flags and warning signs have emerged relatively quickly this year, sounding more like a klaxon alarm than a seductive whisper.
Reuben Philip, Future Famers Class of 2025, director of product, Clarium, spots two overt red flags.
"The biggest red flag is when someone promises that AI can do everything," Philip told Bellwether League Foundation's Leaders & Luminaries. "AI is genuinely capable of doing a lot, but it's common for that word to be used to overpromise and eventually underdeliver, because we overestimate how easy it is to plug something in and get a reliable output.
"For example, consider AI that's classifying thousands of surgical products into categories that clinicians will use to make decisions," Philip continued. "In one case, the AI labeled a group of electrosurgical needles as 'Rigid Needles' because it pulled a physical property from the item description instead of the clinical function. To a data model, 'rigid' is accurate. To a surgeon, it's meaningless. They think in terms of what the instrument does, not what it's made of. If no one with clinical knowledge is reviewing that output, you end up with a system that looks complete but doesn't actually help the people it was built for.
"The second red flag is when a vendor can't go a level deeper. If someone tells you their tool 'uses AI' but can't explain how it works, how they test the outputs, and what the failure modes look like, be cautious. The best AI tools in supply chain today amplify your team's expertise. It doesn't replace the judgment that comes from years of experience. Treating it as a replacement for professionals who understand the nuances of your operation is where things go wrong."
Ryan Burke, Future Famers Class of 2022, BLF Advisory Council, and vice president, Strategic Sourcing, Pandion Optimization Alliance, points to a red flag that has been waving since the use of personal computers, proprietary and open electronic data interchange (EDI) and the need for supply data standards.
"The most immediate red flag is centered around data quality," Burke asserted. "We've all heard 'garbage in, garbage out,' but poor data in the age of AI can lead to strategic errors in supply chain. Poor data could exacerbate demand planning and forecast inaccuracies causing further disruptions in the supply chain."
Background matters, too, according to Vance Moore, Bellwether Class of 2019.
Moore insists that data integrity, unrealistic expectations and perfection can turn progress from AI sideways.
"The biggest issue is the context of your data," Moore argued. "Don't underestimate the very necessary data curation step to realizing the true value of AI. AI is truly amazing at its ability to 'reason' through unstructured data or contextually different structured data, but it is no substitute for human/machine curation as a foundation for success. Many think we will deploy AI and it will find and fix our problems. It may find the issues more quickly than in the past but blind trust is dangerous. Bottom line: You need a continued data curation effort with or without AI as a focus. Data is the currency of supply chain credibility and contribution — invest in it."
Moore cautions against unrealistic expectations. "Set realistic expectations so that you stand a chance at success," he advised. "AI is here; it is not coming. You need to embrace it and implement it with a 'no-going back' expectation. Doing so will take time to build the right foundation. Staged expectations for results is key to progressive confidence. AI is truly a force multiplier and provides remarkable depth of decision insight. The more you embrace the concept and set realistic expectations the sooner you will eliminate false starts or dips in confidence." Moore couches perfection as the enemy that must be vanquished. "Your data is never going to be perfect; the tool will never perfectly do what the marketing pitch projected; you will never have enough time to implement with precision; your team will never fully embrace something that has superhuman capabilities, but all of these imperfect ingredients can be combined to generate something truly special," he noted. "The bottom line: If you wait for perfection, you will become irrelevant and lost in the past. Get started — it will be messy but rewarding once you consume your collective efforts."
Tina Vatanka Murphy, CEO, Global Healthcare Exchange LLC (GHX), and BLF Platinum Sustaining and Educational Partner, understands the appeal but urges caution.
"AI has tremendous potential to transform the healthcare supply chain, but leaders should watch for a few important warning signs," she said.
"One warning sign is AI that adds technology but fails to simplify the work," Murphy explained. Supply chain teams are already managing too much manual work across disconnected systems. If AI becomes one more dashboard or another recommendation to interpret, it will simply add more friction instead of reducing it. The goal should never be to digitize broken processes. It should be to reduce workflow debt and help teams act faster and with more confidence.
"Leaders should also be cautious when AI is built on incomplete context," she continued. "Supply chain decisions connect to clinical needs, financial performance, supplier reliability, inventory availability and patient care. If the data is fragmented or siloed, the
recommendation may appear useful but miss something critical. Teams need to understand not only what AI is recommending, but why.
"Another warning sign is over-automation. AI can help teams spot disruption earlier, identify risk faster and coordinate response more effectively, but it should not remove human judgment from decisions that require clinical, operational or ethical context. The strongest use cases are those where AI augments expertise and supports better decisions," she indicated.
Murphy expresses importance to pay attention to how Tina Vatanka Murphy people are experiencing the change. "If employees feel AI is being done to them rather than built with them, adoption will stall. People need to understand the purpose, trust the output and see how AI helps them do more meaningful work," she said.
"Responsible AI requires more than deploying new technology. It requires the discipline to simplify work, the humility to keep humans in the loop and the courage to redesign processes that no longer serve caregivers, healthcare organizations or the patients who depend on them," Murphy added.
Seeing red blocking out green, black
John Grande, senior vice president and general manager, Spend Management Services & Delivery, Vizient, BLF Founding Sustaining and Educational Partner, calls for context.
"One major red flag is when AI is treated like a feature instead of an operating model," he observed. "Healthcare does not have a shortage of dashboards, alerts or disconnected technologies. The real problem is that value leaks across dozens of operational steps between the contract and the P&L. If AI is not helping orchestrate workflows, improve compliance, automate execution or reduce operational fragmentation, it is unlikely to create meaningful impact.
"Another warning sign is when organizations focus only on productivity gains instead of performance outcomes," Grande continued. "Faster reports and automated tasks are helpful, but the bigger opportunity is closing the performance gap: Reducing leakage, improving price accuracy, increasing contract compliance and helping providers capture more value from every dollar of non-labor spend."
Grande further urges caution about "point solutions that operate in isolation. Supply chain decisions do not happen in a vacuum. Clinical, financial, operational and procurement workflows are deeply connected. AI that is not integrated across those environments can actually create more fragmentation and blind spots instead of less."
Quantity and capability aren't necessarily the tactics for success, according to Grande.
"The organizations that will lead are not the ones deploying the most AI tools," he noted. "They are the ones building more connected, intelligent and orchestrated supply chain ecosystems."
Cody Fisher, president, Concordance Innovations (Surgence), and BLF event partner, also concentrates on AI's corporate context.
"A major red flag is when AI is positioned as an overlay instead of part of an operating system," he observed. "Healthcare has already seen what happens when insight is separated from execution. We get dashboards, alerts and recommendations, but the work itself does not change. Decisions are still delayed. Teams are still chasing answers. Outcomes are still inconsistent.
"Another warning sign is when the system assumes full autonomy too early," Fisher cautioned. "Healthcare supply chains do not tolerate black-box decisions. They require transparency, governance and trust. AI has to earn its role in the system over time, starting with recommendation and decision support, then moving into more automated execution as confidence builds.
"A third is when the solution ignores the multi-enterprise nature of the problem," he concluded. "Supply chain in healthcare is not a single organization problem. It is a coordination problem across providers, distributors, suppliers and partners. Any AI approach that is designed for a silo will fail to deliver sustained value. The common theme across all of these is that technology alone does not fix the problem. The system the technology operates within matters just as much."
Tom Redding, executive vice president, Healthcare, St. Onge Co., and BLF Silver Sustaining Partner, identifies several key red flags and warning signs leaders should watch for when evaluating AI in healthcare supply chain operations.
"We'll fix the process later."
"One of the most dangerous assumptions is that AI can compensate for broken workflows," he observed. "Leaders must first understand and address real operational challenges before layering in automation. Without that foundation, AI doesn't reduce risk, it amplifies it, increasing the potential for failure at scale."
There's a lack of defined baseline metrics."
"Too often, organizations assume AI will 'solve the problem' without first understanding current performance," Redding continued. "Without a clear baseline, it becomes impossible to measure improvement or even know whether performance is getting better or worse. Operating this way is like running a business in quicksand: There's motion, but no traction."
"No change management strategy."
"Implementation is rarely the hardest part, sustaining change is. AI does not eliminate the human element of organizational performance. Without a deliberate change-management approach, even the best tools fail to gain adoption and organizations revert to old behaviors," he said.
Fred Crans, Bellwether Class of 2020, Board Outreach Ambassador, and healthcare business development executive, St. Onge Co., questions three issues.
- Data quality and quantity. "Poor data makes AI output useless."
- The need to construct proper queries. "Failure to ask the right questions produces useless results.
- "Failure to review AI output can miss errors and lead to total lack of confidence in the results, as well as embarrassment (or worse)," he warned.
Charlie Miceli, Bellwether Class of 2023, casts shade on three areas, too.
- "Statements from AI suppliers [that] staffing won't be impacted.
- "General ROI projections made without detailed analysis and good data.
- "If the underlying data isn't good, don't pursue until data is solid," he advised.
Maria Hames, partner, HealthCare Links, and BLF event partner, cites "poor data, lack of engagement (adoption matters as much as the technology), continuous oversight and check ROI. [AI] should improve and not create complexity. Is it solving a problem that you are sure you have?" she indicated.
Meanwhile, Barbara Strain, Bellwether Class of 2021, BLF Board Chairman, and principal, Barbara Strain Consulting, poses three notable questions worth addressing:
- "What processes need to be improved for efficiency?
- "What signals need to be created and automated to prevent errors?
- "How can AI help my staff interact with their customers?"