Artificial intelligence (AI) has developed rapidly, significantly transforming industries, augmenting human capabilities, streamlining operations, and enabling new opportunities. Contemporary AI models excel at automating complex workflows, processing large quantities of data, and facilitating informed decision-making across various sectors. Nonetheless, AI continues to encounter challenges in advanced reasoning and emotional intelligence. To examine AI's strengths and limitations in greater detail, we interviewed Madhu Reddiboina, Founder & CEO of RediMinds, Inc.  This article presents part 1 of that discussion.

Q: How do AI systems struggle with reasoning, abstraction, or common sense?

A: Today’s models (including GPT 5.0 released in August 2025) are brilliant pattern matchers, not grounded thinkers. They reason well inside their training distribution, but struggle with multi-step cause-and-effect situations or edge cases or unstated assumptions. In IRO work, that means an LLM can summarize records fast, but it still needs a reviewer to validate medical nuance and apply policy intent.

 

Q: Can AI truly understand or just simulate understanding?

A: Today it simulates understanding extremely well especially the frontier models when you see deep research in play. For operations, that’s fine—as long as we measure calibration, require citations, and give humans the final say. We must ensure humans remain accountable for values and judgment.

 

Q: What are the limitations of AI in replicating human creativity or emotional intelligence?

A: AI can remix patterns at scale, but it doesn’t feel, care, or build trust in a way that human interaction does. In the IRO world, empathy with providers and patients matters—explaining rationale respectfully in my view is still a human function.

 

Q: Most impressive things AI can do today that weren’t possible five years ago?

A: Reliable long-document parsing, multi-modal intake (PDFs, images, speech), strong coding co-pilots, and agentic workflows that chain tools together. For IROs: auto-triage cases, extract structured facts from messy records, align to policy, and draft rationale with citations—minutes, not hours. In the past work that an assistant or a para legal did can be done comfortably by an AI Agent.

 

Q: Examples of AI outperforming humans in specific tasks?

A: Reading 1,000 pages without fatigue, finding contradictions across files, detecting duplicate/linked claims, normalizing codes, and recalling policy language instantly. But know that humans still outperform on judgment, exceptions, and communication.

 

Q: Which industries are seeing the most transformative impact right now?

A: Software development, customer support, legal ops/contracts, marketing, finance/risk, and healthcare administration. Anywhere there’s heavy text, rules, and repetitive activities are being transformed by AI agents right now.

Q: Surprising or lesser-known applications?

A: Live policy Q&A with verifiable citations; automated prior-auth checklists; credentialing and quality checks; privacy-preserving data linking for outcomes studies etc. are some examples to start.

 

Q: How is AI augmenting human decision-making rather than replacing it?

A: There are several examples of this, Co-pilots like cursor, windsurf are augmenting the engineers to build systems in a fraction of time. Tools like Notebook LM can augment humans to grasp complex concepts fast. All frontier models at the moment are primarily in the business of augmenting humans by answering questions in a meaningful way. In the world of review services, they can pre-summarize records, highlight missing evidence, propose options with pros/cons, and surface the exact policy lines. The human reviewer confirms, amends, and signs—speed + consistency without losing accountability.

 

Q: What capabilities do you expect AI to gain in the next 5-10 years?

A: We can expect:

• More reliable reasoning with fewer hallucinations
• Much longer context windows
• Continuous learning from approved, real-world data
• On-device, PHI-secure AI systems
• Real-time evidence validation and decision support

AI will increasingly handle most administrative workflows end-to-end, with humans supervising exceptions and ensuring accountability.

Q: Any breakthroughs on the horizon that could expand what AI can do?

A: Hyperscale data centers powering the AI boom are already facing energy constraints — a challenge likely to spark major innovation in sustainable energy production. Parallel advances in quantum computing could dramatically accelerate AI reasoning and optimization. On the software side, multi-agent orchestration and secure data enclaves will mature, expanding AI’s role in regulated industries like healthcare while improving trust and compliance.

 

In summary, while artificial intelligence has made remarkable strides in automating complex tasks, parsing vast amounts of data, and augmenting human decision-making, it remains fundamentally limited in areas requiring deep reasoning, creativity, and emotional intelligence. AI excels at pattern recognition and efficiency, transforming industries such as healthcare, legal operations, and customer support. However, true understanding, empathy, and nuanced judgment continue to be uniquely human strengths. As AI evolves, its role will increasingly be to empower humans—handling routine workflows and surfacing insights—while people retain accountability for values, exceptions, and trust. The future promises even greater reliability, longer context windows, and secure, real-time decision support, but the partnership between human expertise and AI will remain essential for responsible and impactful progress.

This document presents part 2 of a dialogue between the NAIRO AI Committee and an AI expert, Madhu Reddiboina, Founder & CEO of RediMinds, Inc .  Artificial intelligence (AI) technologies are rapidly transforming industries, offering unprecedented opportunities alongside complex risks. This expert discussion by the NAIRO AI Committee examines the complex challenges involved in implementing AI, including ongoing problems like model hallucinations, lack of real-world experience, and concerns about bias and over-reliance in important fields such as healthcare.  The discussion highlights best practices for integrating AI with human judgment, balancing accuracy, speed, and cost, and emphasizes the importance of robust governance and continuous oversight.

Q: Why do AI models still hallucinate or generate incorrect information?
A: Their fundamental feature is to predict “likely” next word, not verified truth. Without grounding to a trusted source like a policy database, clinical guideline, claim files etc., they can confidently make things up. We can control the hallucinations with retrieval (always cite), tool use (calculators, policy checkers), confidence thresholds, and a hard prompt to NOT guess or else :).

 

Q: How does the lack of real-world experience affect AI's decision-making capabilities?
A: Clearly AI Models do not have lived experience, so they miss context that humans take for granted such as tradeoffs, edge-case ethics, operational constraints etc. In utilization review or IDR, that shows up as brittle decisions when documentation is messy. We fix it with domain ontologies, scenario training, and ALWAYS keep a human in the loop.

 

Q: How can AI systems unintentionally reinforce bias or discrimination?
A: AI learns from history that is documented and digitized; if history is biased, outputs can be too. Proxies like zip code, benefit type, or provider specialty can encode disparities. We could mitigate some of it via subgroup audits, fairness metrics, bias bounties, and governance that allows humans to override and report harms.

 

Q: What are the risks of over-reliance on AI in critical sectors like healthcare?
A: Over reliance on AI can lead to Automation bias, silent failure, and data drift. In critical workflows, that can delay care or misapply policies. One way to mitigate this is to establish tiered risk controls, clear fallbacks to humans, full audit trails, and continuous monitoring.

 

Q: Best practices for designing AI that works well with humans

A: The approach depends on the process being designed. I start with a process flow or decision map that clearly defines who decides what, and when. Next, identify which parts can be handled effectively by AI given the available data, and which require uniquely human judgment. Then, design an end-to-end workflow that integrates both capabilities. Treat it as an iterative cycle — build, execute, learn, and rebuild — continuously refining how humans and AI complement each other. 

 

Q: Trade-offs between accuracy, speed, and cost in deployment

A: There’s an additional dimension to consider — scale. Speed, accuracy, and scale each influence deployment cost differently. The right balance depends on the primary business outcome you’re optimizing for. If accuracy is paramount, technical and operational choices will differ from when speed or scalability is the goal. Each dimension carries trade-offs in infrastructure, human oversight, and model complexity. Above all, never fully automate high-risk or high-impact decisions.

 

Q: What role will AI play in solving global challenges like healthcare? 

A: AI’s role is to reduce friction, accelerate evidence-based decisions, and expand equitable access to quality care. In the IRO and IDR domains, it enables faster, fairer, and more consistent reviews — freeing clinicians to focus on clinical judgment and human connection. The end result: improved efficiency, transparency, and trust across the healthcare ecosystem. 

 

In summary, responsible use of AI in healthcare requires balancing innovation with strong oversight. With clear governance, transparency, and ongoing human input, organizations can leverage AI to improve clinical decisions, enhance efficiency, and promote fairness in clinical reviews while mitigating risks. As AI continues to evolve, its success will depend not only on technical advancements, but also on a steadfast commitment to ethical collaboration—ensuring that AI augments, rather than replaces, clinical judgment and supports equitable outcomes for patients and providers alike.  

In 1995, Bill Gates appeared on The Late Show with David Letterman to explain this strange new thing called …. THE INTERNET.

Most people did not understand.  Letterman joked, “Why would I want to read baseball scores on a computer when I have the radio?”

AI is increasingly being adopted into UR to streamline review processes, improve efficiency, and potentially reduce costs. AI tools and agents can automate tasks, analyze patterns, perform deep research, and provide decision support, but they may also introduce potential risks like over-reliance and algorithmic bias. NAIRO has observed a dramatic increase in states creating bills around the use of AI in UR. We have noticed that states have taken a wide-ranging approach to regulating AI into their respective UR regulations, but all agree that some form of human oversight is needed. For example, these bills are currently in process:

• Illinois House Bill (HB) 35 (2025) proposes to create the Artificial Intelligence Systems Use in Health Insurance Act, which provides the Department of Insurance with regulatory oversight of health insurance coverage including oversight of the use of AI systems or predictive models to make or support adverse consumer outcomes.
• Maryland HB 820 (2025) Requires that certain carriers, pharmacy benefits managers, and private review agents ensure that AI, algorithm, or other software tools are used in a certain manner when used for conducting UR.
• New York Assembly Bill 8556 (2025-2026) prescribes requirements and safeguards for the use of an AI, algorithm, or other software tool for the purpose of UR for health and accident insurance.
• Rhode Island Senate Bill (SB) 13 (2025-2026) - Use of Artificial Intelligence by Health Insurers, which promotes transparency and accountability in the use of AI by health insurers to manage coverage and claims.
• Tennessee SB 1261 (2025-2026) Insurance Agents and Policies, which imposes requirements for health insurance issuers using AI, algorithms, or other software for UR or utilization management functions.

NAIRO believes there are important benefits in incorporating AI in UR, including increased efficiency and speed. For example, AI can automate routine tasks like data extraction, prior authorization requests, and initial case reviews, freeing up human reviewers for more complex cases. It can also enhance accuracy and consistency. Foundation AI models such as Google’s Gemini 2.5, Open AI’s GPT 4.5 and others now have the context and reasoning capabilities. They can analyze large datasets and identify hidden patterns that could easily be missed by humans, leading to more accurate and consistent decisions. It would result in improved decision support by providing evidence-based recommendations and highlighting relevant clinical information to facilitate human reviewers in making informed decisions.

Reviewer anonymity is an ongoing issue in the independent review industry. The current industry standard for independent review organizations (IROs) is to prevent the disclosure of the reviewer’s identity unless required by contract, statute, or law. The rationale for maintaining anonymity centers on maintaining the overall quality and integrity of the independent review and maintaining the safety of the reviewer. By allowing the reviewer to remain anonymous, NAIRO believes that reviewers are shielded from undue influence and potential harm. Anonymity allows a reviewer to make the most appropriate decision without outside pressure from other review stakeholders. The National Association of Insurance Commissioners (NAIC) Uniform Health Carrier External Review Model Act (Model Act) and several states have enacted specific language protecting reviewer’s identities and their ability to make unfettered decisions. For example, Section 14 of the Model Act holds IROs harmless for their decisions¹. Recently, the Alaska Legislature released Bulletin B 25-05 protecting the identity of IRO reviewers pursuant to AS 21.06.060(f) and (g).

Consumer advocates and treating providers have countered that disclosing a reviewer’s identity allows for additional transparency in the review process. Recently, many treating or attending providers have disclosed reviewers’ identities to stakeholders, including to the insured. Treating providers can access the reviewer’s name from peer-to-peer conversation. In the external review process, the insured is informed of the IRO assigned to their case but does not have access to the reviewer’s name. Information about the reviewer’s qualifications and experience is shared as part of the review decision. In some cases, there have been reported incidents of threats or harassment against reviewers, including actions against reviewers’ medical licenses. As a result of this trend, qualified reviewers may decline to participate in the independent review process, which could severely limit reviewer availability, access, and the overall scope of reviewer coverage from both a geographic and specialty perspective. To offset this potential hazard, reviewers may demand an increased rate of payment to continue to provide IRO reviews. The potential scarcity of reviewers could also force IROs to raise their rates to ensure they can provide appropriate specialty coverage of all review types. Since the insurer pays for state sponsored external reviews in most circumstances, they may pass on the increased cost directly to the insured by raising premiums and related fees.

Imagine a world where technology and Artificial Intelligence (AI) are seamlessly integrated into our daily lives, revolutionizing our experiences in unprecedented ways. Entrepreneur Steve Brown, upon being diagnosed with a rare form of cancer, took charge of his own care team and developed an AI-powered platform aimed at enhancing medical care. This initiative turned his personal challenge into a movement for improved healthcare solutions. One of the agents—an AI oncologist named "Dr. Haddad"—identified previously overlooked patterns, resulting in not only improved insight into his condition but also the establishment of a new pathway for care that holds potential to impact numerous lives.

What began as Brown's mission for survival has evolved into a broader initiative focused on providing individuals with rare and challenging-to-diagnose conditions the opportunity for answers, healing, and hope. His platform, CureWise.com, is already creating a waitlist of patients and clinicians who envision a future enhanced by technology and driven by empathy.

California Senate Bill (SB) 636 could jeopardize the vitality and integrity of the independent utilization review process, and the National Association of Independent Review Organizations (NAIRO) stands firmly against the bill’s provisions pertaining to “duty of care” mandates.

The current version of SB 636 – “Workers’ Compensation: Utilization Review,” introduced by Senator Dave Cortese – would “require employers establishing a medical treatment utilization review process to ensure that utilization review physicians have the same duty of care to an employee as a treating physician.”

Technology solutions are fast-emerging in the clinical peer review space, from workflow tools that can streamline the medical review process to systems like natural language processing and machine learning that can unleash layers of efficiency.

While artificial intelligence-fueled technology, such as ChatGPT and Microsoft’s Bing chatbot, has grabbed headlines in recent weeks, it’s fair to wonder if AI is entering the medical claims review arena. Although it may be a tantalizing thought, experts say it’s not quite there yet.

While a full-fledged vision of artificial intelligence (AI) remains an elusive conquest within the healthcare industry, rapid advances in technology are showing the ability to improve workflow, enable faster decision-making and, overall, deliver a boost to medical review companies and affiliated stakeholders.

When it comes to new-age technology in healthcare, AI is still in a “science fiction” phase, says Ed Bolton, CEO and president of Nexus, a URAC-accredited independent clinical review and utilization management company based in Schertz, Texas, and immediate past president of the National Association of Independent Review Organizations (NAIRO).

Understanding the Vital Role, Challenges, and Opportunities of Independent Medical Review Services

CLICK HERE FOR THE PDF VERSION

Table of Contents

I. Introduction

Effective January 1, 2022, the No Surprises Act (NSA) ushered in sweeping changes to the dispute-resolution process between healthcare providers and payers, establishing a “baseball-style” arbitration system in which an independent arbiter settles payment differences for out-of-network (OON) charges.

The federal independent dispute resolution (IDR) process, which generally applies to group health plans, health insurance issuers offering group or individual health insurance coverage and Federal Employees Health Benefits (FEHB) carriers, includes the certification of IDR entities to make payment determinations.

As ransomware and other types of cyber crime grow increasingly prevalent, it is paramount that organizations in the medical review and utilization review space know how to best protect their business and client operations with adequate levels of cyber-liability insurance.

A growing area of coverage – yet one that can prove challenging to obtain or afford – cyber-liability insurance doesn’t prevent ransomware attacks and data breaches from occurring, but it provides a high level of defense against downstream risks. Most cyber-liability policies provide network security and privacy liability, limited protections against network business interruptions, media liability provisions and limited coverage of legal expenses.

A federal effort to confront escalating mental health issues – including a refocus on mental health coverage parity – was a central part of President Biden’s State of the Union address that he delivered to lawmakers March 2.

The United States is facing what the White House calls an “unprecedented mental health crisis.” Accelerated by the COVID-19 pandemic and the resulting worry, isolation and depression, mental health issues today impact every two out of five adults and an increasing number of children and adolescents.

A mental health parity law that took effect in California on January 1, 2021, means that independent review organizations (IRO) and utilization review organizations (URO) can expect to work with new care guidelines when assessing coverage standards.

Senate Bill (SB) 855 is one of several pieces of legislation to appear in recent months aimed at improving mental health coverage and addressing perceived disparities in the breadth and extent of coverage. Similar laws, including Illinois House Bill 2595, require that health insurers adhere to standards of care developed by nonprofit organizations, rather than the more traditional commercial standards.

The Illinois bill, for instance, requires that “an insurer shall exclusively apply the criteria and guidelines set forth in the most recent versions of the treatment criteria developed by the nonprofit professional association for the relevant clinical specialty.”

With voters in five states passing ballot measures to approve the use of medical or recreational marijuana, the 2020 election cycle shows that the trend of relaxing laws surrounding cannabis continues to gain traction.

 

Fair IDR: The Way Forward To Protect Patients & Resolve Surprise Medical Bill Disputes

Protecting patients from surprise medical bills is a national concern.¹ The stories of financial hardship placed on patients from surprise medical bills, whether due to lack of coordination in our health care delivery system or misaligned billing practices of health care organizations, are well told. In fact, over 57% of American adults have received at least one surprise medical bill.²

Independent medical peer review services are a fundamental part of the U.S. health care system, providing valuable, evidence-based decisions on medical claims.

However, there is an emerging trend that is pushing for faster medical review determinations, particularly in the event of non-acute and non-emergent situations. This trend can potentially compromise the quality of independent review services for the intended beneficiary – the enrollees.

The act of surprise billing – when a patient receives an unexpected medical bill, sometimes for thousands of dollars or more – is a common occurrence within the American health care system. Studies show that roughly 20% of surgical patients receive a surprise bill, at an average of $2,000 in non-covered fees.

Though states are responding differently to the COVID-19 public health emergency (PHE), current stay at home orders in many regions and the hesitation for companies and employees to return to the office have brought attention to a previously under-the-radar element of business continuity. Namely, the potential for operational disruptions brought on by situations where your workforce cannot attend the office, or where your office is forced to close.

Accreditation and certification organizations have varying requirements to respond to the disruption of normal business, whether it is called business continuity, as in the case of URAC, PCI, or HITRUST; Disaster Management (NCQA); or Business Continuity Management (ISO). Emergency situations that can trigger a WFH business continuity model might be anything from the current PHE to prolonged power outages, severe weather, earthquake damage or other disruptive situations.

ALBANY, NY – February 28, 2020, – The National Association of Independent Review Organizations (NAIRO), the nation’s leading voice of independent medical peer review, is pleased to announce its annual Educational Symposium on April 28-30, 2020 at the Kiawah Island Golf Resort in South Carolina.

The NAIRO Symposium offers two days of networking and in-depth education, with sessions focusing on a range of topics relevant to the independent review of healthcare claims.  The event’s 2020 theme is “Protecting Review Integrity During Healthcare and Regulatory Evolution,” and will highlight changes in the healthcare and regulatory environment and their subsequent impact on Independent Review Organizations, Workers’ Compensation Organizations and Health Utilization Management Organizations.