AI is popping up everywhere. There seems to be no limit to what this technology can do or where it can be applied. According to a Gartner 2019 CIO Survey, 37 percent of organizations have adopted AI in some form while an Accenture report claims AI can boost productivity by as much as 40 percent and the global AI market is expected to grow to $60 billion in the next five years. The reasons for the fast adoption rate are largely economic.
Yet, despite all AI can do for businesses, many people see AI in a different light. People are wary of placing too much trust in a machine. Are self-driving cars safe? Do devices with voice command options secretly record conversations?
Workers also have concerns about AI. Will AI make my job easier? Or will AI take away my job?
While some fears are unfounded, others are not. According to a new Brookings Institution report, roughly 36 million American workers will be negatively impacted by AI in a significant way. These 36 million jobs, according to the report, are deemed as having a “high exposure” to automation, meaning that at least 70 percent of the tasks associated with those jobs could be taken over by automation.
With so much at stake, decisions will have to be made. There’s no denying that AI is pervading every industry and affecting workers around the world. So, as a business leader, how do you embrace the future without alienating your workforce? How do you build trust in your AI implementations? Pointing out the benefits to the business is not the best way to bring AI into your enterprise. Enterprises wanting to implement AI solutions need to realize that AI is a journey, not a destination.
Adopting AI Is a Process: Learn to Enjoy the Journey
AI is not just another shrink-wrapped product, and building trust in AI has little to do with AI itself; rather, building trust comes from the things surrounding it. To gain the trust of your employees and to gauge whether you are ready to start implementing AI-enabled solutions, consider the following questions:
- Do we see AI adoption as a journey or a destination?
- Are we committing to do everything needed to be an AI company? This might need changing organization structure, data strategy, processes and personnel.
- Have we built the infrastructure to train our employees to use and develop AI solutions?
- What are our slowest moving areas in terms of new technology adoption? And where have our previous initiatives failed?
- What is our messaging strategy to our organization? And how does it inspire confidence and inclusivity?
- Can we commit to spend time with the solution provider and iterate on the product?
- Are we tolerant to product failures and unpredictability in the early stages of use?
Answering these questions will show which areas in your enterprise need to be reexamined before committing to AI.
Implementing “Human” AI Solutions: Leveraging Empathetic Design
Nothing is more frustrating than trying to do a job with the wrong tools or with tools that were poorly designed or implemented. If someone is already uneasy about AI, a bad experience using an AI-based solution will only confirm that there’s something to fear. This is why you should ensure that any AI solutions that you implement were created using the principles of empathetic design.
Empathetic design, which is sometimes referred to as “human-centered” design, is different from other design strategies in that the goal is to experience the interaction with the device from the point of view of the one using the device. To build trust, the app not only has to work correctly, but it also has to be easy to use and even enjoyable to use. If a new AI-based tool helps people get the job done, helps make their job easier, and helps them be more efficient, their trust in AI will grow. The perception of AI as something to fear will start to change. That’s why it’s so critical to get it right, to make sure the person using the solution has a good experience.
Changing a person’s perception isn’t always easy, but it is possible. I’ve watched a lot of companies going through the process of implementing industrial-grade AI products and the lesson I’ve learned is that for a successful implementation, management and employee engagement is critical, even more so than with general-purpose software products. In my experience, I would say that a successful AI implementation requires as much as 60 to 70 percent more company engagement than what’s required for a traditional automation solution.
It should be noted that even with such a heavy engagement model as this, there will still be some level of uncertainty and fear, regardless of how well the AI solution performs. Nevertheless, this type of engagement, or “co-design,” is critical even for implementations that are ready to go right out of the box.
For industrial AI applications, model validation in the field is a must and is part of the empathetic design strategy. Testing model performance in the real world accomplishes two things: First, it increases the model robustness in the face of unseen challenges; and second, it creates an extra touchpoint for management/worker engagement.
Many might think that this heavy engagement model doesn’t belong in a world that favors a self-service model. After all, automation and self-service models are what drive enterprise software adoption. Unfortunately, this approach has yet to gain traction in the AI space, and I believe this will remain the case for the foreseeable future. This is why AI solutions face an introspection that is unparalleled as to other technologies, and why you need to make sure the AI solutions work and are embraced.
If You Want It Done Right…Creating Your Own AI Solutions
Finding the right AI solution isn’t like shopping for software. For companies who want to create their own solutions, AI has a steep learning curve, requiring an extensive background in both math and science. While someone may be able to use simplified AI solutions without extensive training, to build value by creating AI applications requires a team of experts.
To solve the problem of limited resources, various companies have emerged that boast easy-to-use AI solutions. They claim to have platforms that are so easy to use that anyone can create applications that provide value. Many of these platforms are yet to live up to their hype. And simply adopting a platform is not enough to magically transform your business into an AI-based company.
Having said that, there are steps you can take toward creating a custom AI solution that isn’t just wishful thinking. It starts with empathetic design.
Start by engaging your workforce as a co-designer, and then document exactly what they expect the AI solution to do. Co-design or participatory design is an empathetic design process that involves all stakeholders—workforce, management, provider, and so on. This will make the initial design cycle longer, but it will serve as a great tool to engage all stakeholders. This enables people to build an emotional connection with the product from early stages.
As with more out-of-the-box solutions, here is a list of questions to help you understand where you are in the design process:
- Is our AI product inclusive? Did our design process include both users and sponsors?
- How many touchpoints do we have with our employees during product development?
- How much do we trust the input data?
- Did we build a relationship with the people who will be using this tool?
- Are our inferences suggestive or prescriptive? And how are we validating our inferences?
- Is our AI product improving the day-to-day life of our employees?
- Is our AI product empowering our employees to make better use of their time?
- Does our AI product make our employees more skillful in their task?
Making AI Work for Everyone
Embracing the future of AI doesn’t need to be good for business and bad for employees, or something to be feared. With the proper mindset and commitment to both the needs of the business and the needs of your employees, your company can begin the process of implementing AI solutions that will benefit everyone. When we are using AI to improve the tasks a human does, we need to ensure our product can be trusted, and it can only be accomplished by following a rigorous and empathetic design process.