Generative AI (ChatGPT, Copilot, Gemini): Capabilities, Limits and Risks for Leaders and Managers

1. Insight

Generative AI is now the most widely used form of AI in the workplace. Tools such as ChatGPT, Microsoft Copilot and Google Gemini are already embedded in everyday workflows — through browsers, mobile apps, and increasingly within office software and search engines.

This technology is not experimental or “coming soon”. It is already shaping how work gets done — and it is here to stay.

At an individual level, Generative AI can support a wide range of everyday tasks: drafting documents, summarising information, brainstorming ideas, planning travel, or producing first drafts at speed. In a business context, the potential applications are even broader — from supporting strategy development and communications, to analysing information, designing content, writing code, and accelerating decision-making.

For leaders and managers, the challenge is rarely enthusiasm or resistance. It is understanding what Generative AI can do well, its limitations, how people are accessing it, and what risks and controls need to be actively managed.

Used well, Generative AI becomes a practical productivity accelerator. Used poorly or without oversight, it can introduce avoidable risk, inconsistency and poor decision-making.

2. Why This Matters

Whether you are aware of it or not, your staff are already using Generative AI. In most cases, they are doing so with good intentions — to work faster, improve quality, or reduce effort.

However, responsibility for AI use ultimately sits with organisational leadership. Leaders and managers must ensure that:

• the right tools are being used,

• for the right purposes,

• with appropriate controls, oversight and governance in place.

Unmanaged use creates risk. Managed use creates advantage. The difference is leadership, clarity and intent.

3. Generative AI Fundamentals

Most modern Generative AI tools are built using Large Language Models (LLMs), trained on vast volumes of text, code, images and other data. These models do not “think” or “understand” in a human sense. Instead, they identify patterns in data and generate outputs that appear similar to high-quality human-created content.

Their apparent intelligence comes from scale, training and design — not awareness or judgement.

In practice, Generative AI outputs are shaped by user provided 'prompts' - structured instructions that guide what the model produces, how it responds and the level of detail or style applied.

3.1 Generative AI Capabilities - Practical Use Case Examples

Generative AI is best understood by the types of outputs it can create:

3.1.1 Text Generation

Drafting emails, reports, policies, summaries, proposals and plans. Use Case Example: A programme manager uses Generative AI to produce a first draft of a project update for senior stakeholders. By providing a short prompt outlining key milestones, risks and decisions required, the AI generates a structured and professional briefing note. The manager then reviews, refines and personalises the content before sending it.

Value Delivered:

• Reduces drafting time significantly

• Improves consistency and clarity

• Allows leaders to focus on judgement, not formatting

Typical Applications:

• Executive summaries

• Policy and governance documents

• Business cases and funding proposals

• Meeting minutes and action summaries

3.1.2 Image Generation

Creating illustrations, concepts, diagrams and marketing visuals.

Use Case Example: A leadership team preparing a strategy presentation asks Generative AI to create a simple visual schematic illustrating their target operating model. The AI produces a clean, professional diagram that can be refined and branded, removing the need for specialist design support at the early concept stage.

Value Delivered:

• Accelerates idea visualisation

• Lowers dependency on external design resources

• Enables rapid iteration of concepts

Typical Applications:

• Strategy and operating model diagrams

• Presentation visuals

• Marketing and communication concepts

• Training and learning materials

3.1.3 Audio and Video Generation

Producing voiceovers, transcripts, summaries and synthetic media.

Use Case Example: After a leadership town hall, Generative AI is used to automatically transcribe the session, summarise key messages, and produce a short written briefing for staff who could not attend. In some cases, an AI-generated voiceover is created for internal training materials to ensure consistency of messaging.

Value Delivered:

• Improves accessibility and inclusivity

• Reduces time spent manually transcribing and summarising

• Extends the value of live events and meetings

Typical Applications:

• Meeting and workshop transcripts

• Executive message summaries

• Internal training videos

• Knowledge capture and reuse

3.1.4 Code Generation

Writing, reviewing and explaining software code and scripts.

Use Case Example: A business analyst with limited coding experience uses Generative AI to create a simple data extraction script to automate a recurring report. The AI not only generates the code, but also explains what it does in plain English, allowing the analyst to validate, adapt and safely deploy it with appropriate oversight.

Value Delivered:

• Accelerates automation and productivity

• Lowers technical barriers for non-developers

• Improves code quality through explanation and review

Typical Applications:

• Simple automation scripts

• Spreadsheet macros and formulas

• Code review and documentation

• Learning and capability uplift for teams

Across all capability areas, the pattern is consistent: Generative AI is most effective when it is used to accelerate first drafts, reduce manual effort, and support human decision-making — not replace it.

The real value comes from pairing Generative AI outputs with clear intent, strong governance and active human oversight.

3.2 Benefits of Generative AI - Summary

When used well, Generative AI can deliver clear benefits:

• Efficiency – significant time savings on drafting, summarising and repetitive tasks.

• Quality – more consistent outputs and fewer basic errors when used with review.

• Creativity – rapid idea generation and alternative perspectives.

• Reduced “blank page” effect – accelerating momentum by producing strong first drafts.
  

3.3 When to Use Generative AI

Generative AI is most effective when:

• the task is generative (drafting, ideation, summarisation),

• outputs can be iterated and refined,

• there is a clear need for human review and accountability.

It is far less suitable for tasks requiring absolute accuracy, real-time data, or unverified decision-making without oversight.

3.4 Limits of Generative AI

Even where Generative AI is used in appropriate scenarios, leaders and managers should understand its inherent limitations and the controls required to use it safely:

• Human input and review are essential - Effective use requires good prompting and active human oversight.

• It cannot learn independently - Models do not update their understanding unless explicitly retrained.

• Accuracy is not guaranteed - Outputs can be incomplete, outdated or incorrect due to:

  • limitations in training data,

  • a fixed training cut-off date,

  • inherent model constraints.

Where accuracy matters, outputs should be verified against trusted sources, organisational records, or subject matter expertise. Generative AI should support judgement — not replace it.

3.5 How to Access and Use Generative AI

Leaders and managers should understand the main access routes their teams are using:

• ChatGPT: Available via web and mobile app, with free and paid tiers offering enhanced capability. Enterprise versions provide organisational controls and data protections.

• Microsoft Copilot: Accessible via the web, mobile apps and embedded directly within Microsoft tools such as Word, Excel, Outlook and Teams under appropriate licensing.

• Google Gemini: Available via web and mobile, with integration into Google Workspace tools such as Docs, Sheets and Gmail, alongside enterprise options.

Capabilities evolve rapidly, with new features and products released frequently. This makes clear guidance and governance essential.

4. Key Risks

Without active leadership, Generative AI introduces several organisational risks:

• Uncontrolled (shadow) use outside approved tools and processes

• Lack of awareness and education, particularly around data handling and residency

• Responsible use failures, including misuse or over-reliance

• Weak or absent governance

• Uncoordinated use or poor use-case selection

• Insufficient training, leading to sub-optimal outcomes

• Failure to adopt, resulting in lost productivity and competitive disadvantage
  

5. Mitigating Actions

Leaders and managers can reduce risk and unlock value by taking proportionate action, including:

• setting clear Generative AI usage principles,

• defining approved tools and access routes,

• establishing governance and accountability,

• training teams to use AI effectively and responsibly,

• aligning AI use to real business outcomes.
  

This is where Orr Consulting supports organisations — helping leaders move from ad-hoc use to controlled, value-driven adoption through clear strategy, practical governance and disciplined delivery.

Explore Our AI Services

6. Final Thoughts

Generative AI is not a future capability — it is already reshaping how work happens. The question for leaders is no longer whether it will be used, but how well it will be used.

In the next post, the focus will shift to Generative AI - Maximising Benefits with Prompt Engineering, a simple, practical framework for creating high-quality prompts that dramatically improve Generative AI outputs and can be applied immediately by leaders, managers and teams.

7. Call to Action

If you want to take control of Generative AI use in your organisation — reducing risk while unlocking real productivity gains — now is the time to act.

Explore how Orr Consulting can help you turn Generative AI from an unmanaged risk into a disciplined capability that delivers measurable outcomes.

Request an AI Consultation

Subscribe to Orr Consulting to receive occasional emails with practical AI Insights and updates.

Subscribe to Orr Consulting