The 3 Axioms To Crafting A Human-Centred AI Governance Framework

This article will help you craft an AI governance framework that harmoniously blends the pinnacle of human ingenuity with the forefront of AI technology.

?

Introduction

Imagine you are a newly minted blacksmith. You trained as an apprentice for years, studied your craft for long days and longer nights and you've finally proven yourself to be adept at your craft.

?

You make the pilgrimage to some faraway region to find a blacksmith shop to take up your vocation and a new village to settle in.

?

When you arrive, you're given the role of 'Town Blacksmith' only to find you are spending your time doing everything but blacksmithing. You're writing letters to customers, finding the right suppliers of iron, steel, leather and other materials. You are also delegated some kind of geo-political position because you are now a key voice in the overall village decision-making.

?

What started as a vocation to create tools, contraptions and armor becomes a never-ending busyness to serve the people of your community in every way possible. And, if you somehow find time, actually do some blacksmithing.

?

Sounding familiar? The same reality is faced by practically every knowledge worker on the Earth today. The role you play goes way beyond your title, schedule and, most often, your capacity.

?

A study conducted by the Harvard Business Review found that knowledge workers, on average, spend up to two-thirds of their working time on non-value-added activities such as desk work and “managing across” the organization (for example, meetings with people in other departments) [1]. And these same knowledge workers spend an average of 41% of their week on discretionary activities that offer little personal satisfaction and could be handled competently by others [2].

?

This is, of course, not particularly the choice of most knowledge workers, there's just no better alternative. “My team is understaffed and under-skilled, so my calendar is a nightmare and I get pulled into many more meetings than I should,” one study subject reported. Another commented, “I face the constraint of the working capacity of the people I delegate to." [3]

?

What if your work could return to a state of vocational fulfillment? Imagine choosing work that not only demands your critical thinking and creativity but also produces meaningful outcomes instead of mundane tasks.

?

This choice is the essence of a human-centred AI governance framework, focusing on meaningful, impactful tasks that reinforce the value of our roles and fulfill our need for meaningful work by using AI to augment each labourer rather than automate the processes they are a part of.

?

When I say AI governance, I am talking about the principles, policies, and practices that guide the ethical, responsible, and effective implementation of AI technologies within organizations.

?

To explain it simply, imagine AI as a new team member who needs guidance on how to behave, what tasks to perform, and how to align their actions with the organization's goals.

?

AI governance is the process of setting up the rules for this "team member," ensuring it works well with humans, protects their interests, and enhances overall performance without overstepping ethical boundaries or making decisions that could harm the organization or society.

?

Creating a framework that adopts this philosophy is about putting the human need for meaning at the heart of how your business operates. From the warehouse clerk to the CEO, we all want to do the same thing: work on things that are meaningful to us and others.

?

So, who will do the mundane? The rote and the repetitive? How will we continue to get the tiny, incremental, ever-draining, absolutely required tasks completed without chewing away everyone's most valuable commodity, time?

The Framework:

?

Finding The Pain Points:

?

It's hard to imagine an AI framework that does not focus on the realization of the benefits of AI to individuals from the start [4]. But how do you actually identify that?

?

I'm sure you've come across some of the stock examples like an email writer or a bot that schedules things for you or sends reminders to teammates but at the core of all of these scenarios is the same principle:

?

In knowledge work, when you agree to a new commitment, whether it’s a small task or a big project, it brings with it a certain amount of ongoing administrative overhead. This overhead produces a compounding tax that chews at more and more of an individual's mental bandwidth as their responsibilities grow [5].

?

This can look like being stuck in meetings explaining procedures and policies to new hires or gathering disparate data from departmental silos to create executive reports. It can even just be facilitating an effective strategy meeting focused on creating new lines of business to pursue in the next fiscal year.

?

In all of these cases and many others similar, it is the drain of your mental bandwidth from cognitively intensive tasks that do not drive business value but absolutely need to get done.

?

These are the areas to find within your organization because they provide the greatest chance to drive a real, tangible impact in using AI to change the way your workforce operates.

?

Assessing these pain points not only provides a chance to increase the productivity of your employees but also creates an opportunity to improve the work they do. When you've identified the leeches on their bandwidth and what eats at their day, you're freeing them up to do the actual deep, important work they do with greater focus, intention and quality which has been proven to provide numerous individual, departmental and organizational benefits.

For example, when Lotta Laitinen, a manager at If, a Scandinavian insurance company (https://www.if-insurance.com/about-if/about-us.), jettisoned meetings and administrative tasks in order to spend more time supporting her team, it led to a 5% increase in sales by her unit over a three-week period. [6]

?

Re-Inventing The Process:

?

It's not just about identifying the pain points but correcting them too. When implementing AI systems, you cannot just overlay a model on top of a process that wasn't working in the first place.

?

Identifying, ideating and reasoning on what needs to be changed within the process is largely where the distinction of using AI comes from. When you are able to identify exactly where the bottlenecks are in your process, you can effectively explore how AI can relieve those points of contention.

?

For example, say you are the manager of a small ERP consulting company, your consultants require your approval of all expenses they submit on their timesheet. Of course, you're very busy with important things like creating RFP responses, putting together reports for leadership and planning customer events for the next quarter so the time you can put towards reviewing the consultants' expenditures is limited.

?

The bottleneck, in this case, would be the need to review all line items of the expenses spreadsheet when you're really just looking for any that require you to clarify with the submitter.

?

So, rather than having an AI model that is instructed to examine each line item of the spreadsheet, the AI should be instructed to simply identify any items that do not clearly align with the organization’s policies and procedures. This way, anything important that requires your review is flagged and rather than chewing up bandwidth looking at everything, you only have to examine a few key items and make sure they can be approved.

?

The main idea is you have to identify how the process can be improved before considering what role the AI model will play. Often, the challenge with implementing AI effectively is not necessarily an issue with the model but the situation it is being asked to assist with.

This also provides the chance to understand the cause of the inefficiency and how that can be addressed at a larger level within the organization. You may find that what seemed like a tiny inefficiency in how expense paperwork is handled may be part of some larger inconsistency with company policies and procedures.

?

Reinforcing The Human In The Loop:

?

Central to this framework is the belief that AI should amplify, not replace, human intelligence. By taking over rote tasks, AI allows humans to excel in areas requiring critical thinking, discernment and creative work.

?

Of course, this offers numerous benefits to individuals and their organizations, but this does not mean that individuals can be removed completely from the task. Not only for dependency, safety and ethical reasons but for the sake of learning and development, both for the AI models deployed and the individuals they work with.

?

For example, radiologists leverage computer vision (a type of machine learning) to identify cancerous cells at a higher level of accuracy than humans, but they are still the decision maker as to what operations, treatments and medication to offer to patients for obvious ethical and legal reasons.

?

But this processor-decision-maker relationship has ancillary benefits. Radiologists can learn what AI models are able to see that makes them able to identify problems with higher precision and the radiologists, of course, continue to provide feedback and evaluate AI inferences so that the AI model continues to improve its accuracy further and further as well.

?

What many struggle to realize with AI, and for that matter, any emerging technology, is that the journey of practically implementing the tool is an iterative process. The machine learns from human feedback and humans learn how to use the machine by trying it out and using it in constructed situations.

?

This, of course, only occurs when the humans remain in the loop. When things are completely automated, there is no chance for continuous improvement. The model receives very little constructive feedback because the decision makers are not engaged with its application and this lack of engagement also keeps the decision makers from improving their own understanding of the process outcomes which may be critical for other, higher-value work that an AI cannot do well.

?

When the focus of the AI implementation is maximizing the impact of the human in the loop rather than how to replace them, the organization, the individuals, and the machines that have begun to enter their ecosystem all benefit in an equitable and more prominent way. The sum of all parts is always greater than the value of any individual component.

?

The Framework In Action

?

Now that we've grasped what is important in creating a framework that maximizes the benefits AI brings to your organization, let's see what this looks like in action.

?

Here's a little story [7]:

?

A few years ago, an unnamed Fortune 500 company decided to adopt an earlier version of OpenAI's ChatGPT.

?

This company provides other companies with administrative software. Think like programs that help businesses do accounting and logistics. A big part of this company's job is helping its customers, mostly small businesses, with technical support.

?

The company's customer support agents are based primarily in the Philippines, but also in the United States and other countries. And they spend their days helping small businesses tackle various kinds of technical problems with their software. Think like, "Why am I getting this error message?" or, "Help! I can't log in!"

?

Instead of talking to their customers on the phone, these customer service agents mostly communicate with them through online chat windows. These troubleshooting sessions can be quite long. The average conversation between the agents and customers lasts about 40 minutes. Agents need to know the ins and outs of their company's software, how to solve problems, and how to deal with sometimes irate customers. It's a stressful job, and there's high turnover. In the broader customer service industry, up to?60 percent [8]?of reps quit each year.

?

Facing such high turnover rates, this software company was spending a lot of time and money training new staffers. And so, in late 2020, it decided to begin using an AI system to help its constantly churning customer support staff get better at their jobs faster. The company's goal was to improve the performance of their workers, not replace them.

?

Now, when the agents look at their computer screens, they don't only see a chat window with their customers. They also see another chat window with an AI chatbot, which is there to help them more effectively assist customers in real-time. It advises them on what to potentially write to customers and also provides them with links to internal company information to help them more quickly find solutions to their customers' technical problems.

?

This interactive chatbot was trained by reading through a ton of previous conversations between reps and customers. It has recognized word patterns in these conversations, identifying key phrases and common problems facing customers and how to solve them. Because the company tracks which conversations leave its customers satisfied, the AI chatbot also knows formulas that often lead to success. Think, like, interactions that customers give a 5-star rating. "I'm so sorry you're frustrated with error message 504. All you have to do is restart your computer and then press CTRL-ALT-SHIFT. Have a blessed day!"

?

After the software company adopted AI, the average customer support representative became, on average, 14 percent more productive. They were able to resolve more customer issues per hour. That's huge. The company's workforce is now much faster and more effective. They're also, apparently, happier. Turnover has gone down, especially among new hires.

?

Not only that, but the company’s customers also are more satisfied. They give higher ratings to support staff. They also generally seem to be nicer in their conversations and are less likely to ask to speak to an agent's supervisor.

?

Takeaways

?

There are many moving parts to consider in any AI governance framework. There is no one size fits all and organizations need to find what works for them, their workforce and their broader leadership network.

?

However, there are some foundational elements to consider, adhering to these axioms will give your organization the best chance of applying an intuitive, robust and battle-tested AI governance framework that puts your workforce at the center of each layer and drives meaningful change in how individuals approach and do their best work.

?

Core Axioms of Human-Centred AI Governance Framework

?

1.???? Identifying and Alleviating Pain Points

• Proactive Identification: Start with a proactive approach to identify areas where cognitive overload and administrative burdens undermine employee productivity and satisfaction. It's about recognizing the tasks that consume disproportionate mental bandwidth without driving direct business value.
• Tangible Impact: By pinpointing and addressing these pain points, AI governance aims to free up employees to engage in more meaningful, fulfilling work that leverages their unique human capabilities, thereby enhancing both individual well-being and organizational performance.

2.???? Re-inventing Processes with AI

• Process Optimization Before Automation: Before AI integration, there's a critical step of re-evaluating and re-engineering existing processes. The focus is on understanding and fixing the underlying inefficiencies within a process to ensure that AI implementation is both effective and enhances the workflow rather than simply automating a flawed process.
• Strategic AI Deployment: The deployment of AI must be strategic, focusing on relieving bottlenecks and enhancing decision-making processes. AI should be tailored to flag critical issues for human review, ensuring that efforts are concentrated on areas requiring human insight, thus optimizing time and resources.

3.???? Reinforcing the Human in the Loop

• Amplification of Human Intelligence: Central to this framework is the principle that AI should serve to amplify, not replace, human intelligence. AI implementations should empower humans to excel in tasks that require critical thinking, discernment, and creative problem-solving.
• Iterative Learning and Improvement: The framework emphasizes a symbiotic relationship between AI and humans, where both learn from each other. Humans provide the necessary feedback for AI to improve its accuracy and effectiveness, while AI insights contribute to human learning and decision-making processes. This iterative loop ensures continuous improvement and adaptation.
• Preservation of Human Oversight: Ensuring that humans remain integral to the decision-making process, not just for ethical and safety reasons but also as a means of fostering learning and development for both the AI models and the human operators. This mutual evolution maximizes the benefits of AI across the organization, making the whole system more than the sum of its parts.

???

These axioms embody a holistic approach to AI governance that prioritizes human well-being, process efficiency, and the strategic use of technology to enhance, rather than replace, human capabilities. The focus is on creating a framework that is not only technologically advanced but also deeply human-centric, fostering an environment where both humans and AI can thrive and contribute to the greater organizational good.

?

Have any questions or wondering how to get started with implementing AI the right way in your organization? Feel free to shoot me an email: [email protected] or message me directly on Linkedin to further assess your situation!

?

References:

?

[1,2,3,6] Make Time for the Work That Matters by Julian Birkinshaw and Jordan Cohen: https://hbr.org/2013/09/make-time-for-the-work-that-matters

?

[4] Human-centricity in AI governance: A systemic approach (Anton Sigfrids, Jaana Leikas, Henrikki Salo-P?ntinen, Emmi Koskimies) https://www.frontiersin.org/articles/10.3389/frai.2023.976887/full#:~:text=Researchers%20applying%20the%20concept%20emphasize,2021%3B%20Shneiderman%2C%202022

?

[5] Slow Productivity – Cal Newport, Chapter 3: Do Fewer Things

?

?

[7] This company adopted AI. Here's what happened to its human workers by Greg Rosalsky: https://www.npr.org/sections/money/2023/05/02/1172791281/this-company-adopted-ai-heres-what-happened-to-its-human-workers

[8]Customer care: The future talent factory https://click.nl.npr.org/?qs=407c71695417f37ac193b94ff23368891eaf4113e4d4ddf78f97b0affb28e7886b3a89fd13f1c58114cc34896e4d34ac2886e1db52386213