The discussion around artificial intelligence often centers on replacement. Will AI take our jobs? The real shift, however, may not be about machines replacing people, but rather empowering them. AI offers tools to make us more capable, productive, and valuable in our roles. The leaders who succeed in this new era will not chase every new tool. Instead, they will learn how to wisely adopt AI and guide their teams through these changes.
Tim Staton and Kate Marshall recently discussed these critical topics. Kate Marshall is the founder of The Gray, a fractional chief AI officer, and author of “AI at Work.” She helps organizations move beyond AI hype. Her focus is on practical, responsible adoption that delivers real outcomes. With a background in cybersecurity, workforce enablement, and AI strategy, Kate Marshall offers a grounded perspective. She shows leaders how to innovate while keeping the human element central. Their conversation covered AI literacy, responsible adoption, and leading in a world shaped by intelligent technology.
Many organizations remain in an experimental phase with AI. They want to scale up, bringing in agents and automations. Yet, they face challenges. These include legacy systems and, more importantly, human fear. Employees worry that AI agents might replace them. This fear leads to many pilot projects not reaching full-scale implementation. Roughly 70-80% of piloted projects fail to complete. The core issues are often data, people, and legacy systems.
The narrative around AI frequently highlights cost reduction through workforce reduction. Companies announce layoffs, and their stock prices rise. This creates a perception that AI’s primary goal is to cut jobs. However, a more productive conversation would focus on how AI can supplement workloads. AI can remove redundant tasks. This frees up time for employees to drive exponential growth. As Kate Marshall points out, this should be the message. Unfortunately, movies and news stories often fuel the fear of job displacement. She believes AI will bring massive abundance. Transparency about AI goals and benefits is crucial.
AI is not yet ready to completely replace humans. Many frustrating experiences with AI customer service bots confirm this. The real value lies in collaboration. Businesses that thrive develop an “exoskeleton” mentality. Human orchestrators guide AI agents. They assign tasks and develop automations for repetitive processes. The human then reviews the AI’s output, applies judgment, and ensures goals are met.
Kate Marshall stresses that she does not fully trust AI to execute autonomously. Human involvement remains essential. While “unicorn-style” organizations with one CEO managing many AI agents might emerge, human direction is still necessary. The power comes from human orchestrators communicating with various agents and automations. This hybrid approach allows businesses to leverage AI’s strengths while maintaining human oversight.
Successfully onboarding AI as a strategy is a significant challenge. Many organizations simply update their AI policy and approve tools, then wonder why things fail three months later. The issue often lies in neglecting the human aspect. It is not enough to bring in tools. Employees must be empowered to learn and use them effectively. Kate Marshall describes this as giving people “keys to this fancy car” without teaching them to drive.
When working with companies, Kate Marshall helps them choose the right large language model (LLM), such as Google Gemini, Claude, or ChatGPT. She then assists in training the model on the company’s data, business processes, and people. Crucially, she trains the employees on how to orchestrate and interact with the AI for maximum benefit.
A major barrier to AI adoption is concern over data integrity. Organizations worry about where their data resides, who accesses it, and if competitors might gain access. AI models learn from data, raising significant privacy and security questions.
Kate Marshall advises against using free AI plans. For any serious use, a paid plan (e.g., $20/month LLM plan) or an enterprise solution is necessary. Free plans often use user data for training. Different systems suit different priorities. Healthcare providers, for instance, must adhere to HIPAA regulations. Financial institutions handle personally identifiable information (PII). These sectors require specialized solutions that prevent sensitive data from interacting with standard LLMs.
Specific considerations for data security:
These adoption policies are often customized. They depend on the tech stack and industry, making a one-size-fits-all solution difficult.
Leaders should not let data integrity concerns discourage them. The goal is to think critically about data’s journey: where it goes, who keeps it, and how its integrity is maintained. While cybersecurity emphasizes maximum security (e.g., a laptop at the bottom of the ocean), this approach hinders progress. Leaders must balance risk with the potential for exponential growth.
Kate Marshall is forward-leaning on AI usage. She believes leaders must accept risks in a smart way. Building guardrails is essential, but organizations must innovate. They need to try new tools. Stepping back completely is not an option for thriving businesses.
Many users initially find LLMs (Claude, ChatGPT, Gemini, Grok) unhelpful. They describe the output as “AI slop” or generic content. This happens when the AI is not trained and customized. To get better results, users must personalize the AI.
To train an LLM effectively:
A well-trained AI can produce output that sounds uniquely yours. This takes some effort but dramatically improves results, making the AI feel less generic.
When an organization introduces AI, training is vital. Different employees will have varying levels of AI understanding. Kate Marshall recommends running multiple sessions tailored to different baselines.
Providing dedicated time and space for learning is crucial. Many existing tools, like Slack, already have embedded AI features (e.g., Slack bots) that go unused due to a lack of training. Organizations only use a fraction of their tools’ capacity. Investing in continuous training helps employees understand how to manage, build with, and use these tools effectively. Relying on “University of YouTube” for self-learning is an option, but organizational support is key.
Understanding “human in the loop” is critical for responsible AI deployment. This concept differentiates between automations and AI agents.
Regardless of whether it’s an automation or an AI agent, “human in the loop” means a physical human reviews the process. They offer judgment, confirm correctness, and provide approval at various checkpoints. Without this human oversight, agents can “go rogue” or produce “AI slop.” Human orchestration prevents errors and ensures the AI stays on the right path. Keeping humans involved in both automations and agents remains vital.
Selecting the best AI tool for an organization begins with the goal, not the tool itself. Kate Marshall emphasizes: “You always want to start with the goal and not the tool.” First, define what you are trying to achieve or build. Then, work backward to determine which tools align with that objective.
The evaluation process includes:
For individuals and organizations, clarity on “where you want to go” and “what you want these automations and tools to do” guides the selection process. Then, assess if you have the components to build it.
“AI at Work”: A Practical Guide
Kate Marshall wrote “AI at Work” to address the practical needs of her clients. She noticed a consistent need for checklists and structured steps, regardless of business size or focus. The book is a workbook, not a novel. It guides non-technical professionals on how to start implementing AI.
The book covers:
The guide is regimented and customizable, providing checkbox items for personal and professional AI integration. It offers a “start here” approach for anyone new to AI.
The conversation about AI agents and automations raises questions about organizational preparedness. How will AI agents integrate with existing employees? How can organizations prepare their workforce for this? These questions prompt deeper thought:
The traditional organizational chart will likely look very different in 3-4 years. Few organizations are fully prepared for this shift. The future will be “wild and weird,” but also exciting. Organizations must consider what the world will look like in 4-5 years. Then, they need to backtrack and identify current priorities to manage and support people through this transition. The emergence of fractional chief AI officers suggests a growing recognition of this need. These roles collaborate with HR, executive teams, and IT to drive change. However, comprehensive examples of organizations “phenomenally well ahead of the curve” are still rare.
The AI revolution is not about replacing humans but augmenting their capabilities. By focusing on responsible adoption, effective training, and a “human in the loop” approach, organizations can leverage AI to boost productivity and unlock new growth. Addressing data integrity and security concerns, while balancing risk with innovation, are critical steps. Leaders must prioritize understanding and integrating AI strategically, always starting with their goals rather than just chasing new tools. This thoughtful approach ensures that AI becomes a powerful partner in the workplace.
Curious to know more about Tim Stating The Obvious? Contact me through the contact form below to delve deeper into the world of leadership excellence!