UnitedHealth’s AI Plan: Redemption Through Innovation

•
May 15, 2025    

After a disastrous AI rollout that denied Medicare claims with a 90% error rate, UnitedHealth Group isn’t retreating from technology – it’s doubling down with a bold new vision.

See Also: OnDemand | Navigate the threat of AI-powered cyberattacks

A 2023 class-action lawsuit criticized the company for its nH Predict AI tool, developed by NaviHealth, which had a 90% error rate, denying Medicare Advantage claims and leaving patients confused. The accusations were damning: an AI meant to streamline decisions created chaos and distrust. Adding to the crisis was the U.S. Department of Justice’s inquiry into Medicare billing practices, potentially tied to AI-driven diagnostics. The final blow came with the tragic assassination of UnitedHealthcare CEO Brian Thompson in December 2024, which caused a social media firestorm. UnitedHealth’s reputation was at stake. Its AI initiatives were labeled reckless. Yet amid the negative sentiment, the company has presented its new AI vision, deploying over 1,000 AI applications and promoting responsible AI. This is UnitedHealth’s high-stakes bid for redemption through innovation.

The company disputes claims that nH Predict is a decision-making tool rather than a decision-guiding one, but the damage is visible. The DOJ probe deepens the scrutiny and raises questions about whether UnitedHealth’s AI-driven diagnostics inflated Medicare billings. The killing of CEO Brian Thompson, while not directly related to decisions made by the AI tool, became a pivotal reason for outrage. UnitedHealth’s AI misstep wasn’t just a legal or PR crisis – it was much more.

To dismiss UnitedHealth’s AI innovations entirely would be unfair. According to The Wall Street Journal, the company claims to have over 1,000 AI use cases across its insurance, healthcare delivery and pharmacy divisions, with another 1,000 in development. This is not surface-level tinkering. The company appears committed to AI-enabled healthcare. AI is directing 26 million consumer calls with precision, cutting wait times and routing inquiries to the right advocates, as former CEO Andrew Witty said in the Q1 2025 earnings call. By year-end, AI agents will handle over half of all calls – a scale few can match. At Optum Insight, AI-powered claims efficiency tools have boosted productivity by over 20%, the company claims. UnitedHealth’s pharmacy services are also being reshaped, with AI optimizing drug pricing and cutting up to 25% of prescription reauthorizations, a move that could save billions by 2028.

Sandeep Dadlani, UnitedHealth’s chief digital and technology officer, calls this approach “pragmatic, ambitious,” and he’s right. The company’s 20,000 engineers – most working on AI – are transcribing clinician visits, powering chatbots and accelerating software development. Optum, the company’s innovation engine, uses AI to provide value-based care, prioritizing outcomes over volume. The 2025 proxy statement lays it out: Health technology is one of the five strategic pillars, with AI and clinical data driving simpler processes, sharper clinical decisions and greater transparency. UnitedHealth isn’t just deploying AI – it’s betting on it. It aims to connect clinics, homes and virtual platforms into a seamless ecosystem. The numbers support the ambition: a 20% reduction in prior authorizations in 2023, a national Gold Card program rewarding evidence-based care and a pledge to pass 100% of pharmacy rebates to clients by 2028. The company refuses to be defined by its past failures.

Ambition alone won’t erase the stain of nH Predict. UnitedHealth knows this, and its responsible AI – RAI – policy is key to its turnaround. Chief AI Officer Rahul Bhotika calls responsible AI a “collective responsibility.” The RAI policy mandates fairness, transparency and safety, with a human-in-the-loop model ensuring AI doesn’t replace clinicians but supports them. UnitedHealth now has a dedicated RAI board that enforces HIPAA-compliant standards, continuous monitoring and bias mitigation. The goal is to address every risk that contributed to the nH Predict debacle. Practical applications demonstrate the policy in action: AI tools help patients find high-quality providers swiftly, while clinicians can use AI-driven risk identification to flag conditions like type 2 diabetes without autonomous decisions. UnitedHealth’s global team of AI experts, along with partners, emphasizes explainability and reliability to ensure systems don’t just function but are trusted. The company’s governance framework is built to withstand scrutiny from regulators, patients and a skeptical public.

UnitedHealth’s focus on responsible AI doesn’t appear to be merely damage control. The nH Predict lawsuit exposed the dangers of unsupervised AI, and UnitedHealth’s response seems like a course correction. By embedding ethical rigor into every layer of its AI strategy, the company is trying to rebuild trust it may have lost. The company’s RAI framework ensures that as UnitedHealth scales its 1,000-plus AI applications, it does so with appropriate guardrails. It’s a calculated move to rebuild credibility in an industry where trust is currency.

This is not an easy task for UnitedHealth. The lawsuit’s fallout still lingers, and the DOJ probe could have major implications for UnitedHealth and the industry at large. The Wall Street Journal reported on Thursday that the DOJ is allegedly investigating UnitedHealth for possible criminal fraud related to its Medicare Advantage business practices, but that UHG said in a statement that it hasn’t been notified by the DOJ about a criminal inventigation.

Scaling AI to handle over half of consumer calls or reducing reauthorizations risks new errors if execution is faulty. Responsible AI is only as strong as its implementation. Critics argue the company’s AI push is too little, too late – a desperate bid to address the trust deficit. But the bold steps taken by learning from its missteps to redefine healthcare’s future can’t be ignored.