The Biggest AI Mistake Most companies are Making

0
170

Share on LinkedIn

Much of AI news focuses on the importance of regulation, data risks, and intellectual property debates. And of course there is the rally cry “AI is coming for our jobs.” In the midst of the-sky-is-falling narrative, there are grains of truth. Last year even we at The Petrova Experience cautioned against artificial intelligence without regulation and shared our thoughts on guiding principles that keep humanity at the center of this innovation.

We also stated, early on in the AI revolution, that applications like ChatGPT, and the growing number of language processing applications, first and foremost, are tools. Tools that like any others, can be leveraged optimally or suboptimally. As such, they can transform the ways we work, the scope of jobs we do, and the manner in which we serve customers.

These opportunities led to publications like AI Magazine covering the concrete ways AI impacts our way of life. I call it AI Magazine since it is also written by AI and is on the topic of AI. If you decide to invest 15 USD in this magazine you will read how various applications of language processing models can help us plan our next trip, leave our aging family members in the company of a “buddy,” and get a mega-smart digital reader of X-Ray results.

The concept of AI as a tool seems straightforward enough. But, in the noise and enthusiasm of most AI news coverage – not to mention the average cocktail party chatter – the binary “sky is falling” vs “sky’s the limit” conversation omits the biggest gap in the understanding of AI for customer experience. This gap is the misuse of artificial intelligence as a new tech functionality vs artificial intelligence as a new IT capability.

Is AI a new Tech Functionality or IT Capability?

By and large, both AI advocates and detractors perceive artificial intelligence as a new tech functionality. What do I mean by that? A functionality is turnkey. It is already designed for the end user. All brands need to do is “turn it on.”

The problem is, AI is far from being “ready” for the end user without a significant amount of work from IT. To get adequate ROI without brand risks like the AI chat that curses (more on that later), organizations need to treat AI as a new IT capability. A capability that needs to be developed in customer-centric use cases.

That brings us back to our AI is a tool argument. Use AI to create a new product, end-user experience, and customer outcomes. Do not think of it as a finished product, experience, or outcome.

Let’s unpack this idea more. A functionality is an output or performance of a system. It is a solution that is complete in itself. When we focus on this definition, it becomes clear: functionality does not accurately describe artificial intelligence.

A capability, on the other hand, is a system’s potential or ability. That includes its potential to solve problems, improve workflows, and increase efficiencies. Now it sounds like we are talking about the solutions to end user problems that LLMs can bring and the new ways of life that computer vision has enabled in the last decade.

Did you know the Google Translate we all use is only possible because of computer vision?  Or that the progress made in autonomous cars could not have happened without computer vision? This is what we are talking about! All the cool things we experience as humans are the functionality that teams across the world are designing, testing, and launching with AI capability. This is the path to transforming employee and customer experience, creating efficiencies, and reducing costs.

Optimizing AI in Your Organization

Recognizing AI as a capability rather than a functionality is the first step to figure out how to optimize it in your organization.

Given AI is a capability, the question becomes, how do we train it to solve our problems and serve our objectives?  To leverage AI as a tool your organization can rely on, start by applying intentional design.

Gather your teams and think through their problems.  Match those challenges with the capabilities of AI to determine what problems this new technology can solve. But remember, do not forget your existing internal capabilities.

Even more importantly, get down to the brass tacks of customer experience design. Bring the voice of your customer into the design of AI-powered projects. Gather and analyze customer research data to ensure you are designing the functionalities your customers or employees need, rather than those you assume they need or want.

Bring expert CX strategy and vision to the AI applications. Before we dive deeper into that ideal state, a word of caution.

AI Gaps When Using AI as a Capability Prematurely

The risk of using a AI as a capability too early is that you miss critical steps.  In this scenario, you neglect to gather thorough requirements, fail to create testing protocols, and leave out robust user experience testing.  Inevitably, this approach creates experience breaks for customers. And clean-up work for employees. In the worst cases, it also yields brand reputational damage.

Customer experience fails and brand reputation threats are common today. This is why we read articles like this, about DPD’s chatbot cursing at a customer.  The chatbot started cursing and criticizing its own company. A system “update” triggered that failure. Back in the day, when we were updating systems less advanced than AI, we scheduled 3-5 weeks of function and end-user testing. I cannot speak on behalf of DPD, but something tells me leaders today are underestimating the need IT teams have to test all permutations of AI-driven applications. We are deceived by the friendly UI. And so we risk forgetting the basics of building complete, safe applications.

This is the result of looking at AI as a plug-and-play functionality rather than the capability it is. A capability you can train and leverage. Language learning models running chatbots are an ideal CX application of the technology tool. They can even serve as a way of moving your customer along the customer journey in a more personalized environment that ultimately increases product utilization and fosters loyalty. But the technology cannot do that on its own.

Leveraging AI effectively starts as all impactful technology applications start: with the quality of the data.

AI Costs and Returns 

If you look at AI as a low investment and a cost saver, you are bound for failure. Yes, it can decrease costs over time. However, like all positive impact solutions, AI implementation requires investment in data; computing power through energy;  and labor costs for data scientists.

Historically, many organizations have underinvested in data cleanliness. Data scientists tend to be the unsung heroes. In the early 2000s, we only heard about data problems when a customer did not receive a financial statement or when a bug in the data caused customer dissatisfaction.  Today, the stakes are much higher. If the data is not clean, the entire AI capability is wrong.

So, why did organizations underinvest in data cleanliness? Because it is expensive. It hits the Operating Expenses (vs the capital costs of platforms and servers), and it is ongoing. Pulling clean, usable data requires significant hours from data scientists – the most expensive IT costs in your stack. The problem with this specific cost is that you need to keep generating ROI from the application you have built to PAY for the data costs. Suddenly, the business case of AI does not look so good to the CFO.

But this investment is precisely what makes the AI tool usable – and valuable. Fed with the right data, AI works for you instead of against you.

The Right AI Mindset

To leverage AI capabilities effectively, start by developing the right AI mindset. Step one, understand artificial intelligence as the capability it is. And the limitations and opportunities it offers. Next, understand there will be significant costs to leverage AI correctly.

To be effective with AI, start small. Design a proof of concept (POC). Leave room to flex into new developments and integrate AI in your existing technologies. Recently, I completed an AI course. The instructor shared that six months ago, she was training on things that do not exist anymore! Transformation and innovation are happening faster than we have ever experienced. That means it is time to learn how to be agile while you are also learning to build facility with the tools.

In fact, agility is the exact mindset you need to leverage AI for your organization. But it’s not the agility we were all talking about 5-10 years ago when we were reshaping organizational structures and fostering digital transformation.

Agility on Steroids

Back then, when we built an application, we used the waterfall approach. Our processes were sequential and clean. IT waited for a few months while the business created its 150-page requirements document. Then, the business waited another 9 – 12 months to see their application show up. If we wanted to change something, we had to defend the change and wait months to see it.

The biggest upside for IT was that costs were predictable. Roadmaps were defined. And, to a great extent, the organizational power was in the hands of IT. The business rarely got involved in the technology side of the house. We worked in silos with predictable outputs, including go live dates.

After the waterfall era, we moved into the Agile Methodology. During the agile phase, we were running sprints in cross-functional teams with scrums and features of the applications coming live every month. We continued to push toward our go-live dates.  Agile helped organizations develop a flexible mindset and expand on the idea that building technology can be a reiterative process.

Looking back, it’s like we were getting ready for AI during that period. Without the Agile project management approach we could not have developed the agile mindset required today.  We could not grasp what is happening now that artificial intelligence capabilities are entering our daily operations. To develop AI we need to think of  a new approach. An approach we call agile on steroids. Successful organizations will have IT teams that look at the AI applications DAILY to manage them properly.

Agile was our prep course for what is coming. And what is coming is a perpetual design and re-design environment. Previously, design started from capabilities. Now, we are introducing the opposite direction, in which we will launch our tech and then observe and manage it to discover the new capabilities that emerge. We will be co-creating with the AI technology. This is a significant mindset shift. And we have been preparing for this for years (whether we knew it or not!).

How Does AI Change the Every-Day?

So, if we are facing agile on steroids, what does it mean for your organization? It means development, exploration, and innovation are never going to stop. Learning and designing new ways to leverage artificial intelligence to bring greater value to employee teams and customers will be part of everyone’s job.

AI maintenance will be a daily task, in perpetuity. We will have IT specialists who are the “buddies” of a language model, “taking care” of it.  And so, as leaders, we need to build the mindset of perpetual learners. And we need to foster the same mindset in our employees who will also become increasingly able to delve into the more creative, generative parts of their jobs. Remember, AI is not replacing employees as much as it is giving employees back time to take ideas to new heights and co-create more valuable solutions that have internal and external impact.

The AI era – the era of agility on steroids – requires a new level of stamina. Leaders who want to know the exact outcomes of their projects or products will not excel in this environment. However, leaders who are flexible and can see the world in perpetual motion with ever-evolving capabilities, will win in the next decade.

How AI Transformation is Like a Car

In case the AI landscape is still a little fuzzy, we want to leave you with an analogy that is working for us and our clients. Imagine you have a car. That car is in your driveway. It is yours, and you will always have it.

Now, think about how you use your car. You can use it to go places. It might take you around the corner to the store, or it might take you to another country. Same car – around the corner, or to another country. Now, think about the different types of cars you can choose. And what they enable you to do. Trucks. Farming. Transportation of goods and services. Autonomous vehicles…

You can go nearly anywhere, do nearly anything, depending on how you calibrate your vehicle (or what “car” you choose). It offers limitless possibilities, much like AI. However, as all drivers know, your car is always going to need something: fuel, tune-ups, replacement parts, etc. If the car is sitting in your driveway not taking you anywhere, you are wasting money and time on it. Its ROI comes from the way you use it. Because, at the end of the day, your car, like AI, is your tool.

Now, think about where you want that tool to take you and your firm. What can you do with AI to improve employee and customer experience? How will end users tell your AI story?

As experience designers, we at The Petrova Experience are the makers of those stories. Through intentional, human-centered design, we facilitate end-user stories that bring you the ROI on AI. To shape your story, spend some time talking with us.

Republished with author's permission from original post.

Liliana Petrova
Liliana Petrova CCXP pioneered a new customer-centric culture that energized more than 15,000 JetBlue employees. Future Travel Experience & Popular Science awarded her for her JFK Lobby redesign & facial recognition program. Committed to creating seamless experiences for customers and greater value for brands, she founded The Petrova Experience, an international customer experience consulting firm that helps brands improve CX. To elevate the industry, she launched a membership program to help CX professionals grow their careers. Ms Petrova lives in Brooklyn with her husband and daughter.

ADD YOUR COMMENT

Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here