How Strategic Implementation Leads to AI Innovation: Fostering Intelligent AI Usage

Global investment in AI has skyrocketed over the past few years, driving a transformational shift in workplace practices across nearly every industry. As organizations rapidly move to adopt AI into their day-to-day workflow, concerns over thoughtful implementation and security are steadily rising.

In fact, studies show that very few companies have any real discipline when it comes to AI implementation. In a KPMG study of 48,000 global workers, 50% said that they used AI tools without knowing whether or not they were allowed, and 44% said that they knowingly used AI improperly at work.

While AI presents a multitude of benefits to workers and companies at large, there seems to be a sort of lawlessness surrounding its usage. How, then, can organizations be innovative with AI while ensuring responsible governance?

Ownership and Oversight: The First Step to Safe AI

While it can feel natural to rush into the new and exciting world of AI, it’s crucial to remember that creating a mature AI policy is a strategic effort. If you want your company’s AI usage to be disciplined and low-risk, you must create a policy that thoughtfully incorporates governance, oversight, regulatory frameworks, and organizational readiness.

Cultivating a culture of mature AI usage in the workplace means proactively considering the risks and taking the proper steps towards mitigation. This alone puts you far ahead of the curve. In a study on AI in the legal industry, Axiom found that only around two fifths of leaders had implemented AI use policies, while a third of leaders hadn’t taken any steps at all toward mitigating risks.

1. Start with ownership. Build out a small group that can work together to create an AI policy that keeps your company’s needs at top of mind. A group like this would ideally include execution leadership, technical know-how (IT/security), policy guidance, and business supervision (HR/finance). This group should also look into establishing cross-functional teams and task forces devoted to AI initiatives.

2. Schedule check-ins. Once you have your ownership group, a quarterly “AI governance roundtable” can serve as an ongoing effort to achieve accountability and auditability.

3. Set up clear roles. Establish clear RACI (Responsible, Accounted, Consulted, Informed) mapping for monitoring, greenlighting, and suspending tools. Consider trackable governance metrics like percentage of approved AI tools, incidents, and pilots reviewed.

Defining Your AI Policy

Once you’ve built your leadership team, it’s time to consider what a mature AI policy might actually look like. This will involve building out a document that covers a myriad of functions within your company’s AI framework.

1. Data and Privacy Policy: This covers data that may or may not be used by AI, how data must be prepared, and retention and deletion rules. It also specifies obligations when handling regulated or privileged materials.

2. Third-Party/Vendor Use Policy: A guide for procurement that sets the criteria and contractual guardrails for any external AI vendor or cloud service: required DPAs, model-training exclusions, retention limits, audit rights, breach-notification terms, and minimum security certifications.

3. Client-Facing Tools and Disclosure Policy: This is guidance that defines when client interactions may involve AI, what disclosures or consents are required, and standardized wording for client-facing disclaimers.

4. Model Testing and Validating Policy: A document prescribing pre-deployment and ongoing checks that all AI tools must pass: accuracy sampling, bias/fairness audits, data loss prevention (DLP) simulation or red-team checks, and acceptable error thresholds. This defines audit cadence, required logging/provenance, and the go/no-go criteria (and rollback triggers) tied to objective KPIs.

5. Employee Acceptable Use Policy: A policy that gives staff a thorough understanding of what is and is not acceptable AI usage when it comes to their day-to-day tasks. It should include a list of approved/banned tools, prompt hygiene basics, and a quick one-page reference.  

Training Teams To Use AI Intelligently

After you’ve established a disciplined framework, you can start bringing the wider team on board. It’s not enough to send a company-wide email establishing the rules. Instead, make sure that your team is equipped to handle AI by teaching them how to use it efficiently, intelligently, and creatively. This works best as a multifaceted approach.

1. Use multiple learning formats. Stave off boredom and phone checking by offering interactive workshops, online modules, and hands-on projects.

2. Listen to different presenters. Exhibit a variety of expertise by bringing in external experts, members of your AI leadership group, and other staff members.

3. Design role-specific training. It can help to use AI training that is tailored to your industry, with scenarios that mirror your company’s everyday workflow.

4. Execute layered delivery. Aim for training that’s upfront (onboarding), ongoing (regular skills development), and as-needed (major updates, new features, or regulatory changes).

5. Apply integrated measurement techniques. Track engagement, retention, and applied outcomes to refine training over time.

6. Consistently push ethics and responsible usage. Weave ethical considerations throughout your training (and not just as a standalone module).

Continuing Governance After Initial Deployment

Once you’ve established your company’s AI framework and trained your team to use it in a responsible and ethical manner, it may seem like the work is done. But policy alone isn’t enough to maintain an environment of intelligent AI usage.

Companies will still need to vet vendors, establish guidelines on when and when not to use generative AI, set memorable rules for AI usage, and decide how to handle hallucinations (hint: treat them as a legal liability). This is why it’s so important to keep humans in the loop for all of your AI output. While policy takes you far, the actual act of using AI in the workplace can still present problems if you’re not careful.

Our whitepaper includes the tips you see here, as well as a more thorough look at AI adoption, the AI tool landscape, common governance challenges, and how to achieve AI maturity. While the setup may seem like quite an involved process, a disciplined approach will save you a lot of trouble down the line. The beauty of establishing a seamless AI workflow is that you’ll spend less time wondering “is this safe?” and more time creating knowledgeable, innovative AI solutions for your everyday challenges.