What companies keep getting wrong about AI implementation

AI rollouts don’t always go as planned. While the technology promises efficiency and innovation, real-world deployments often create new problems — and more human work — instead.

When AI promise meets business reality

Before worrying about AI replacing people, it’s worth examining how it is actually performing in the real world — where automation often creates more work, not less.

Ten years ago, IBM announced with great fanfare that Watson for Oncology was as accurate as human physicians in reading X-rays, CT scans and other reports. In some regions lacking oncologists, IBM even promoted Watson as a potential substitute for doctors.

But the reality soon surfaced. According to ASH Clinical News, internal documents revealed that Watson made unorthodox and unsafe recommendations when provided with synthetic (rather than real) patient data. Ultimately, IBM sold Watson Health’s data and analytics division to a private equity firm in 2021 for $1 billion — after investing more than $5 billion.

IBM wasn’t alone. Remember Zillow Offers?

Zillow built an AI model to predict home values and aggressively bought homes based on those predictions. The algorithm consistently overpaid, leading to half a billion dollars in losses and mass layoffs. The program collapsed in less than a year when the algorithm failed to adjust to a cooling housing market.

Dig deeper: Implementing AI without a problem is a fast road to failure

Even recent rollouts still miss the mark

You might say, “But those examples are years old. Surely companies have learned their lessons by now.”

AI capabilities have indeed improved dramatically in a short time. But the rush to push out AI-powered updates hasn’t slowed down — and not all rollouts are being handled well. Unfortunately, we had a front-row seat to a more recent misstep.

Like many small businesses, we rely on Intuit’s QuickBooks Online to run our operations. Recently, QuickBooks rolled out an AI-powered version of the platform. For us, it’s been nothing short of a disaster.

Here’s what we encountered:

• Forced adoption: Unlike other platforms that let customers opt in or pilot new features, Intuit pushed us into the AI version.
• Faulty machine learning: Although trained on transactions, QuickBooks frequently miscategorized payments based solely on dollar value. If a vendor sent one $1,000 invoice, all invoices for that vendor were recorded as $1,000.
• Coding problems: Payments to contractors were recorded under QuickBooks payment instead of the contractor’s name.
• Hallucinations in accounting: Categories were randomly assigned in ways neither we nor our accountants could explain — or fix.
• Passing costs to customers: The issues became so bad that we had to pay our accountants thousands of dollars to troubleshoot, with no resolution.
• Poor communication: No notice of the change, no documentation and no guidance on how to roll back.
• Broken workflows: Critical functions, such as invoicing, were disrupted. At one point, email addresses dropped off invoices altogether (including mine and the client’s) and emails started getting flagged as spam.

The biggest sin is that QuickBooks sits at the heart of our business. Cash flow, payroll and client billing are all dependent on it. When AI upgrades destabilize that core, the consequences ripple across the organization.

And this isn’t unique to QuickBooks. These examples — IBM Watson, Zillow, Intuit — are reminders that AI implementation is not just about technology. It’s about trust, communication and responsibility.

Dig deeper: Your AI strategy is stuck in the past — here’s how to fix it

Key takeaways for companies rolling out AI

Each example shows AI rollouts fail not because the technology lacks power, but because execution lacks care. These are the principles that can keep innovation from turning into disruption.

• Don’t force change on customers: Allow opt-ins and pilots before mandating a new version.
• Validate in the real world, not just the lab: Test extensively with real customer data and workflows.
• Design a rollback path: Customers need a fast way back if things break.
• Prioritize communication: Explain what’s changing, why and how users should adapt.
• Respect the mission-critical nature of your tool: The more essential the product, the higher the standard for reliability must be.
• Measure downstream impact: An upgrade to AI can affect payments, compliance or customer relationships in ways that go far beyond the software itself.

Building AI that earns trust

AI has the potential to transform industries — but poor implementation can do real damage. The cost of rushing AI into production without testing, communication and accountability isn’t borne by software companies. It falls on the businesses and people who depend on them.

Correcting AI’s mistakes often requires more human work, not less. The real winners won’t be those who ship first, but those who build systems that are reliable, transparent and trustworthy.

Technology should enable businesses. When the intelligence in AI isn’t backed by thoughtful design, it becomes both a technical and a business failure.

The post What companies keep getting wrong about AI implementation appeared first on MarTech.