Artificial intelligence is quickly becoming a competitive necessity, but like any new technology, it faces skepticism and mistrust. Here's how to create effective change management processes to build trust and engage users.
There are three ingredients you need to make artificial intelligence (AI) work for your enterprise: Data + Change Management + Trust.
In my last article, I focused on data — how to make sure your data is ready for AI. Today, I’ll discuss how to create effective change management processes to build trust and engage users.
The challenge of building trust
Humans have a strong tendency to question, mistrust, or simply ignore predictions.
I’ve seen time and time again, organizations add an AI-based prediction or recommendation, only to find that users don’t trust it. When they don’t trust it, they don’t take action and, worse, may dismiss or discount the value of subsequent predictions.
The first step in defining your change management strategy is to ask how AI will impact business processes:
- Are you embedding AI in existing systems & workflows?
- Do those workflows need to change?
- Are you creating entirely new end-user workflows?
Your answer will guide the steps you need to take to educate, train, and support your users.
Some applications are ‘naturally’ suited to prediction-based improvements. For example, if you provide an email marketing manager with predictions on audience engagement (e.g., response rates, click rates, unsubscribe rates) and guidance on actions they can take to improve engagement, it doesn’t change their existing process. Embedding AI in the campaign management workflow has few barriers to adoption.
Other use cases involve far greater change. When AI results in a net new business process, or creates substantial changes to workflows, be ready to invest time to educate end users and prove (and socialize) business value.
For instance, you may have a sales manager accustomed to using Excel for forecasting. Chances are they add (unintentionally or not) their own judgment into the forecast. If you add a predictive forecast into their workflow, that’s a significant shift. Suddenly, machine learning is giving them information on top of what they already know about the state of the pipeline. It may be difficult for them to trust the prediction because it’s asking them to change the way they work.
4 ways to build trust with your users
To pave the way for long-term success, start with these four strategies to encourage your users to engage with and trust the AI you build.
1. Invest in education
Take the time to educate your line-of-business executives and users on what you are doing and the fundamentals of AI itself. Your end users need to understand how AI has the potential to improve their outcomes.
All too often, I see organizations overlook this step in the flurry of excitement that AI can generate. And the resulting decisions made on priorities and budgets can be disastrous.
BCG reached a similar conclusion in their research on scaling AI in the enterprise. "Yet, even as many companies have begun applying AI solutions with impressive results, few have developed full-scale AI capabilities that are systemic and companywide."
Don’t make that mistake.
Education is vital in setting reasonable expectations. It arms the ultimate beneficiaries of the solution with a clear understanding of what they are getting and how.
2. Provide context and transparency
In cases where you show a prediction directly to the user, provide transparency into how the machine arrived at that prediction.
One approach is to show the top predictive factors in your model that led to the prediction. But you need to strike a balance between explaining the prediction and drowning the end user in excessive detail or surfacing obscure, machine-generated factors. Keep it simple: less is more.
3. Make results relatable
Think about how your favorite mobile navigation app plots alternative routes to your destination. The app also estimates the additional time required for those routes. You want to know the impact of choosing a different route.
Build trust with business users by giving them that same level of insight. For example, you can show a salesperson the expected impact of proposing a particular discount tier or deciding the next step in their sales process.
4. Create the opportunity for continuous feedback
Predictions are probabilities, and there are times when the predicted outcome will be wrong. To build trust, create an easy mechanism to give feedback on predictions your user believes to be inaccurate or unhelpful.
This engages your users and has the added benefit of allowing you to improve model accuracy by incorporating feedback into your training datasets.
AI is fast becoming an imperative as senior executives recognize the competitive necessity for investment. Those who fail to embrace AI will quickly find themselves lagging the market as the pioneers race ahead.
Don’t be daunted. Think big, but start small (and soon!) And throughout the process, remember that Data, Change, and Trust are your critical enablers of success with AI.
Discover more about artificial intelligence and how it makes customers happier at our Resource Center.