Capital One, one of the largest US banks, has used artificial intelligence to reduce quality errors in audit risk and compliance by 85%, according to MIT Sloan Management Review. The bank’s intelligent assistant, built using open-weight models and advanced analytics, has also saved 5,000 audit hours across 30 auditors in a few months, with one task cut from five hours to three minutes. AI is also used for fraud reduction, software development, and customer service, including generative AI to help agents answer customer queries more quickly and accurately. Several hundred agents now use AI-powered call summarisation, which has improved productivity and customer experience.

Capital One’s leaders say the bank’s approach to AI is focused on enterprise-wide capabilities, not just individual productivity. Training is prioritised to ensure staff can use new tools effectively. The bank’s chief enterprise AI officer, Prem Natarajan, argues that transformation with AI cannot be measured by short-term ROI, but by long-term customer satisfaction and operational improvement.

For regulated-industry leaders, the Capital One example shows the scale of operational gains possible from AI in core risk and compliance functions. The bank’s focus on training and enterprise adoption may offer a model for other financial institutions and insurers seeking to improve quality and efficiency at scale.

Otto Dettmer/Ikon Images

Imagine you're an operations manager at a global retailer who has invested heavily in data science. Your team produces highly accurate demand forecasts, yet when leadership asks how much inventory to hold and where, the answers are vague. This disconnect between data predictions and actionable business decisions illustrates a common issue: data models often don’t address the real business problems.

Studies show 60-85% of data science projects fail—not due to poor algorithms, but because data scientists and managers often speak different languages. Data scientists focus on prediction accuracy, while managers need clear guidance on decisions that impact costs, profits, and market share. This gap leaves many valuable insights unused.

Decision science offers a solution by modeling the actual business decisions managers face, aligning analytics directly with business objectives. Instead of just predicting outcomes, decision science translates data into actionable strategies, ensuring analytics deliver measurable value.

Data science excels in prediction and machine learning, but decision science emphasizes problem framing, business knowledge, and decision modeling. Combining these disciplines strengthens problem definition and encourages collaboration between analysts and business leaders.

For example, forecasting product demand is useful, but managers must decide production quantities and inventory levels based on profit contribution, costs, and capacity constraints. Decision modeling incorporates these factors and provides recommendations or scenarios that balance risk and return.

Decision science also helps avoid the “convenient data” trap, encouraging teams to gather the data that truly addresses the decision problem, not just what’s easiest to access. This approach fosters metrics aligned with business goals, like financial impact, instead of generic accuracy scores.

To maximize ROI from analytics, organizations should start projects with a clear decision-focused brief, develop enterprise digital twins to simulate business scenarios, insist on actionable recommendations—not just predictions—and build teams blending data science with decision science expertise. Leadership that bridges technical insights with business decisions is essential.

Ultimately, organizations must shift from being data-driven to decision-driven, ensuring analytics translate into better business choices and stronger outcomes.

For the fifth consecutive year, MIT Sloan Management Review and Boston Consulting Group (BCG) have worked with a global panel of AI experts to explore the state of responsible AI implementation. Recent research warns that excessive dependence on AI may weaken our cognitive and ethical judgment, leading to what experts call "epistemic de-skilling." This concern has even spurred measures such as New York City's ban on generative AI in schools to protect children’s development.

When asked if preserving trustworthy human judgment amidst advancing AI is the defining challenge of responsible AI, 65% of the experts agreed, emphasizing that human judgment is a critical skill at risk of erosion but essential to preserve. They argued that human judgment brings context, wisdom, and accountability that AI lacks. However, 19% disagreed, citing the inherent flaws and limitations of human judgment, while 15% remained neutral. The consensus is that the goal isn’t preserving human judgment for its own sake but designing AI systems that enable meaningful human judgment where it matters most.

Experts highlighted that human judgment remains vital for high-stakes decisions, ethical trade-offs, and situations where AI lacks context or encounters novel challenges. But human judgment itself has weaknesses, including fatigue and bias, and simply keeping a human "in the loop" does not guarantee trustworthiness. Overreliance on AI risks displacing human judgment and reducing the skills needed to verify and intervene effectively.

The path forward lies in human-machine collaboration designed to leverage strengths of both. Organizations should identify where human judgment is critical, deliberately build opportunities for humans to exercise judgment, and use training and feedback loops to maintain and improve judgment skills.

Recommendations include defining roles for human judgment at the earliest design stages, creating real opportunities for intervention, practicing judgment regularly, tailoring training by experience, and continuously assessing the effectiveness of human oversight. This careful balance can preserve human agency and accountability while maximizing AI’s value.

In essence, responsible AI means designing systems that keep human judgment strong, trusted, and active—when and where it truly counts.

Jakob Ingebrigtsen, a world-class middle-distance runner, exemplifies how a pacer supports elite performance by managing pace and reducing mental strain. Athletes use pacers to maintain optimal speed and focus on crucial aspects of their performance rather than continuously monitoring their timing. Similarly, leaders often face challenges due to moving too fast or too slow, impacting decision-making, team motivation, and alignment with strategic goals. Many leaders lack a pacer, leaving them vulnerable to pace-related pitfalls that hinder effectiveness.

Research involving 125 employees revealed that 93% observed leaders whose pace didn’t match situational needs, with 75% running too fast and 25% too slow, yet only 40% had a pacer. Pacers help leaders calibrate their speed—slowing down fast leaders to improve deliberation and learning or prompting slower leaders to accelerate decisions and action. They help maintain strategic focus, manage team energy, and foster alignment, reducing burnout and disengagement.

Effective pacers are trustworthy, knowledgeable, and understand the organizational context. They may come from various roles: senior executives, coaches, peers, chiefs of staff, or trusted direct reports. Choosing the right pacer depends on leadership stage and pacing challenges. For rising leaders, mentors or experienced peers serve well, while veteran leaders might benefit from independent board members or coaches.

The best pacers enable leaders to develop internal rhythm and judgment, eventually reducing the need for external pacing. They empower leaders to conserve cognitive resources and energy, focusing on priority decisions and pacing sustainable performance. Proper use of a pacer is about enhancing, not replacing, a leader's self-awareness and accountability, ultimately helping leaders win the race with greater purpose and control.

Alice Mollon/Ikon Images

Recent research involving interviews with senior marketing and digital leaders across multiple sectors reveals critical insights into how companies adapt SEO for AI-driven search. Key findings identify three challenges: gaps in AI visibility metrics, resistance to changing SEO workflows, and the rapid amplification of inaccurate brand information by AI.

A case study of an investment platform, Stakewise, highlights the risks when AI-driven referrals increase but the accuracy of AI responses about the brand falls short, leading to misinformation about products and services.

The rise of generative AI search demands a new discipline called Generative Engine Optimization (GEO), which focuses on two main challenges: ensuring brand visibility in AI responses and maintaining accuracy of that information. While traditional SEO helps AI find relevant content, GEO ensures AI cites and conveys the brand correctly.

Effective GEO requires content to be structured for AI extractability and extends beyond owned websites to third-party reviews and forums. Technical access for AI crawlers and ongoing content updates are vital.

Accuracy remains a major concern, with companies lacking processes to identify and correct AI-generated errors. A disciplined approach involving systematic querying, source tracing, targeted corrections, and ongoing monitoring helps maintain brand integrity in AI answers.

New GEO metrics—share of voice, citation patterns, and AI-referred engagement—help measure AI visibility and accuracy, guiding strategic adjustments. Managers are advised to audit AI perceptions, improve content clarity, manage offsite authority, and measure performance regularly.

In a zero-click AI search environment, brand visibility is no longer just about being found—it’s about being accurately cited and described. This shift demands a comprehensive reputation strategy, making AI-driven brand management a pivotal competitive advantage.

In 2021, Barcelona-based delivery platform Glovo confronted Spain’s Rider Law, which threatened its contractor-based business model by presuming employment when algorithmic systems control couriers. To avoid reclassifying couriers as employees, Glovo redesigned its app to emphasize courier autonomy—allowing free logins, order rejections without penalties, and daily rate multipliers. However, by 2024, this workaround collapsed under legal scrutiny, protests, and financial penalties, compelling Glovo to hire 15,000 couriers as employees, incurring significant costs.

This example illustrates the growing complexity of “algorithmic minimal compliance,” where companies try to meet regulatory requirements with minimal changes to their core model. Although initially appealing, such workarounds can backfire by expanding audit exposure, increasing internal complexity, and intensifying regulatory scrutiny. The study draws from five European platform cases—Glovo, Deliveroo, Airbnb, Uber, and Meta—to identify four red flags that signal when minimal compliance fails: tightening legal clarity, visible social harm, escalating regulatory scrutiny, and growing internal complexity.

The article contrasts Deliveroo’s narrow, contained workaround, which held due to a specific legal test and limited impact, against Glovo’s extensive system redesign that ultimately triggered full employment reclassification. Additional cases demonstrate how durable compliance requires embedding obligations into product design, ensuring meaningful human oversight, and maintaining ongoing regulatory access.

Leaders are advised to rigorously assess regulatory tests, stress-test compliance changes for systemic effects, set clear escalation triggers, build robust evidence for compliance claims, and develop reusable compliance infrastructure. Early and informed decisions can preserve management’s options and avoid costly forced redesigns imposed by accumulated workarounds. These insights apply not only to platform firms but any connected system-driven business facing algorithmic regulation.

Digital tools designed to improve processes frequently fail to deliver expected results, leading to frustration among users and misdiagnosis by managers. Often, failure is attributed to poor adoption or insufficient training without a deeper investigation into the cause. A critical error made in these postmortems is treating data quality and system usability as one issue, when in reality, they fail in different ways and affect users differently.

Research led by Danae Diaz and colleagues surveyed physicians to study the impact of data quality and system design on cognitive load during high-stakes decisions. The findings reveal that good data quality fosters deeper, meaningful engagement (germane cognitive load) necessary for sound judgment, while usable system design reduces unnecessary mental effort caused by poor navigation and clutter (extraneous cognitive load). However, improving one without the other leads to frustration and reduced decision quality.

For example, investing in excellent data without improving system interface can overwhelm users with navigation friction despite better information, while a smooth interface with poor data still results in weak decisions. Proper diagnosis requires separating these issues and addressing both through governance—eliminating data redundancy, showing data lineage, mandating essential fields, and managing alert thresholds appropriately.

This distinction is vital for AI systems as well, where enhancing AI recommendations must be matched with intuitive interface design to maximize benefits. Leadership evaluations should focus beyond adoption and accuracy metrics to questions about cognitive design—does the system aid or hinder finding relevant information? Are alerts actionable or ignored? Are decision makers supported to deliver better outcomes?

Leaders who understand and manage these cognitive design elements will more effectively harness digital investments, avoiding repeated failures and improving decision quality across any knowledge work setting.

In July 2024, a faulty software update from cybersecurity firm CrowdStrike caused a massive 78-minute IT system crash affecting 8.5 million Windows devices worldwide, disrupting airlines, hospitals, and banks, with losses exceeding $5.4 billion. Earlier that year, a ransomware attack on Change Healthcare crippled insurance claim processing nationwide, exposing sensitive data and costing over $2 billion. These incidents highlight a critical question: Had organizations rehearsed such crises effectively? While many enterprises use digital twins—virtual, real-time replicas of their operations—for optimizing processes, few employ them to simulate crisis scenarios. A poll of senior security practitioners revealed that 91% do not use digital twins for crisis simulations, citing leadership awareness as the main barrier.

Traditional tabletop exercises, though useful, have limits: they rely on pre-set scenarios and test teams rather than systems. AI alone offers minimal improvement in anticipating crisis points. Digital twins, when used as continuous rehearsal environments, enable organizations to stress-test real systems against unforeseen disruptions, fostering faster decision-making and deeper operational insight.

This article proposes a five-step framework for repurposing digital twins into crisis rehearsal tools: treating simulations as ongoing practices, positioning AI as a structured sparring partner, embedding anticipation in organizational culture, creating scenario templates for rapid engagement, and adopting a maturity model from reactive to anticipatory crisis readiness. Companies like BMW and Walmart illustrate successful digital twin deployments that enhance operational resilience. Building rehearsal intelligence is no longer a technology challenge but a strategic imperative needing leadership commitment today.

The research involved interviews with 39 executives across 10 countries and diverse industries, plus an online workshop with 36 enterprise data leaders to explore how digital technologies address sustainability challenges. Additionally, surveys of 360 Swiss manufacturing employees revealed approaches to energy efficiency. Rising global energy costs, heightened by geopolitical events like conflicts involving the U.S., Israel, and Iran, press companies financially, especially in energy-intensive sectors such as cement, chemicals, and metals. These industries, along with manufacturers relying on complex supply chains, face profit margin pressures due to increased fuel and power expenses.

Businesses are expanding efforts to enhance energy efficiency not only to support sustainability but also to achieve tangible cost savings, bolster resilience, and enable growth. Some have cut energy consumption by up to half at certain facilities and improved process efficiency by 10-20%. Success stems from comprehensive strategies integrating digital system upgrades and clear executive accountability with performance targets.

Three key strategies emerged: first, investing in data collection and analysis to enable real-time energy management through sensors, dashboards, and centralized platforms; second, redesigning operations to optimize energy use, reduce waste, and deploy AI for continuous improvement; and third, creating products and services that help customers meet energy efficiency goals.

Companies like Clariant utilize digital platforms and AI to monitor and reduce energy use across global sites, achieving significant savings. Similarly, Georg Fischer combines energy-saving technologies, renewable energy, and circular economy practices to lower fossil fuel dependence and emissions. Producers like ABB and DMG Mori offer energy-efficient products and data-driven services that help customers reduce consumption and improve operational efficiency.

The findings highlight a reinforcing cycle where visibility into energy consumption drives operational redesigns, funding further innovations for customers. Organizations that prioritize energy efficiency with clear goals position themselves advantageously amid volatile fossil fuel markets.

At Alturas Capital Partners, advances in generative AI have enabled in-house legal teams to handle lease work traditionally outsourced to outside counsel, delivering significant cost savings and faster turnaround times. This shift threatens some external providers with irrelevance as AI reduces their role, reshaping how firms define their services. While some firms expand their offerings, others lose business to clients adopting AI internally. The pressure on pricing is evident, with major banks recently urging law firms to reduce fees due to AI efficiencies.

Generative AI is lowering costs for producing legal documents, market insights, creative assets, and software, enabling in-house teams to replicate many functions once outsourced. To stay relevant, service providers must not rely solely on expertise but also compete on production cost and ease of delivery.

Providers should focus on three key strategies:

1. Improve Unit Economics: By investing in scalable, automated systems and leveraging proprietary data, firms can produce outputs more cost-effectively than clients can in-house. WPP exemplifies this with its agentic marketing platform and self-serve options that combine data, expertise, and AI tools.

2. Simplify Procurement: Reducing the friction in purchasing services by integrating AI agents that fulfill requests directly within clients’ existing tools makes outsourcing more attractive. Moody’s and law firms like A&O Shearman demonstrate this approach by embedding AI-driven services into clients' workflows.

3. Manage the Operational Burden: Taking on quality control, compliance, and ongoing maintenance costs that clients face if they internalize work strengthens the provider’s value proposition. Thomson Reuters’ CoCounsel Legal product is an example, offering robust, legally vetted AI research that firms cannot easily replicate internally.

Service providers must critically analyze which deliverables are vulnerable to in-house AI production and develop cost-effective, easy-to-access, and low-maintenance offerings. Success lies in making outsourcing cheaper, more convenient, and operationally less burdensome than building internal AI capabilities. While AI challenges traditional expertise, it also raises the bar for providers to innovate and integrate technology with human skills to stay competitive.