Key Considerations When Implementing AI

Bain & Company’s recent study showed that companies with high AI implementation budgets aren’t seeing the expected return on investment. This is what I expected and what I’ve continued to see happen with any new “hyped” technology without a clear goal or need. We explored this in my prior article.

“Nearly 40% of companies that measured AI cost landed below 10% of their expected savings, despite targeting 11% to 20%. Yet 90% are increasing their AI budgets again.”

— Bain & Company’s Automation and AI Pathfinder Survey 2026

Clearly these businesses struggled to understand that they needed a clear problem to solve for AI to enhance efficiency and innovation. However, for the 50-60% of organizations that are experience a higher return, many now face a different critical challenge: How do we implement responsible AI practices while ensuring the speed that continuous improvement requires?

AI will most likely never fully replace humans, but unchecked implementations with no governance can erode the empathy, creativity, and ethics that fuel true innovation. When paired responsibly with human expertise, AI becomes a transformative tool for efficiency toward a clear goal. This power comes with responsibility: leaders must implement ethical governance and structured change management to prevent biased automation while democratizing accurate information.

Below, we break down AI governance considerations and actionable strategies for iterative process improvement—helping businesses scale intelligently while keeping humans at the helm while driving real business growth.

Key Considerations for Ethical AI Integration

AI adoption requires addressing three pillars: bias mitigation, governance, and change management. Without proper safeguards, AI holds a lot of risks, so we at Lini have pulled together recommendations for each of these three pillars that businesses can utilize to best harness AI's efficiency while minimizing those risks.

1. AI Bias Mitigation

AI must be designed to bridge divides, not amplify echo chambers. Without ethical guardrails, AI entrenches biases, monopolizes narratives, and degrades media literacy—a risk underscored by rising misinformation and historical repetition (like resurgent fascism). AI learns from historical data—meaning biased inputs lead to biased outputs. “Garbage in, Garbage out!”

For example, an AI recruitment tool might unintentionally favor certain demographics if trained on skewed hiring data.

Recommendations to Ensure Ethical AI:

Commit to transparency in AI training data and decision-making (sign the Human-Centered AI Manifesto).
Diversify and Audit training datasets for fairness, representation, to counter algorithmic bias and monopolized narratives
Utilize existing tools: like AI bias detection tools (e.g., IBM’s AI Fairness 360).
Prioritize human-centered experiences such as AI should support, not replace, critical decisions and human experiences. True ethical reasoning in ambiguous situations still requires human judgment.

2. AI Governance & Accountability

Storytelling is the bedrock of moral ethics and historical lessons—from campfires to boardrooms. AI dilutes this by replacing human narratives with synthetic content. Without clear governance, accountability and ownership, AI becomes a "black box," eroding trust.

Recommendations to Strengthen AI Governance:

Use AI as a starting point: AI can be a research tool, but keep storytelling and finese human-led (e.g., keynote speeches, culture-building, final editing)
Democratize information access: documenting decision logic, and making sure it shares its sources allows teams to understand how AI arrived at conclusions - especially making sure all understand, not just data scientists.
Develop AI oversight: Assign clear roles for Human-AI collaboration and consider forming a cross-functional AI oversight team (IT, legal, operations) to audit AI outputs.
AI performance reviews: Scheduling regular AI reviews ensures ongoing compliance, not just at implementation.

3. AI Adoption (Change Management)

Keeping AI human-centered isn’t optional at this point—it’s required. AI lacks moral reasoning; humans must steer it for impact. These days, resistance to AI often stems from fear of job displacement or mistrust in machine-led decisions. Change is constant across organizations, so management is built into your processes and implementation, especially around AI changes, allowing organizations to build trust.

Recommendations for Human-Centered Change Strategies:

Early & ongoing feedback: Involve employees early in AI testing and training. Also require human sign-off on high-stakes AI decisions (e.g., hiring, strategy). Encourage experimentation.
Foster a Culture of Learning: Champion AI data and media literacy training programs to demystify tools and upskill teams to work alongside AI tools.
Messaging: Consider clear and transparent messaging when it comes to AI implementation, such as positioning AI as an efficiency booster, not a replacement. Highlight its role in automating repetitive tasks (like data sifting) so employees can focus on creativity, vision-setting, and stakeholder relationships.
Measuring Outputs: Focusing on the digital efficiency and effectiveness gains of AI implementation can only get you so far, adding cultural KPIs (e.g., employee trust, stakeholder feedback) allows for a more holistic picture. Also utilizing the “human element” to validate AI output is correct.
Centralize Data: As AI usage grows consider how AI data will scale responsibly across teams. Consider using centralized insights (e.g., predictive analytics) to empower frontline employees with AI tools tailored to their needs.

Once your organization has considered and planned for these various AI implementation risks, you’re ready to move on to building out your change management and optimization plan.

Be Ready to Lead the Human-AI Alliance

AI won’t replace leaders, but leaders who can harness the power of AI will replace those who cannot. AI should amplify human potential—not replace it. The winning formula combines:

AI’s computational power for efficiency gains.
Human skills like empathy, creativity, and judgment.

At Lini, we help organizations implement AI and any other new emerging technology with responsible frameworks, ensuring technology serves human-centric goals.

Ready to audit your AI readiness? Try our AI Readiness Audit and follow us on LinkedIn or Instagram for more insights on AI and operational efficiency.

If you’d like more reading, please consider reviewing the OECD AI Classification questions and metrics.