Expect some turbulence as GenAI
takes off

Our survey of early adopter superstars provided clear guidance (see Exhibit 1) on their biggest obstacles to deploying generative AI (GenAI). GenAI is ready for prime time, but don’t expect it to work “out of the box.” Get ready to pedal up a steep learning curve in the coming quarters.

Four challenges that require leadership focus and action

The four biggest speed bumps every company needs to roll over are trusting the bots, getting the right humans in the loop, ensuring real return (vs. creating another investment sinkhole), and managing data to ensure the bot can deliver value.

• Learn to trust the bots
  GenAI solutions probably seem as magical to us as the first wireless radios and grainy television images must have seemed to our forebears. We naturally tend to mistrust things we can’t explain, so it’s little wonder that experienced GenAI users rank “inability to explain and govern the AI black box” as the biggest blocker to rolling out GenAI in the enterprise.
• Get the right humans in the loop
  GenAI solutions will not arrive on your company’s doorstep and self-inflate like a bouncy castle. Bringing GenAI solutions “to life” takes an ecosystem of technologies like data platforms, large language model (LLM) engines, massive compute resources, and conversational layers. They all need (and will continue to need) people-that’s us. Early adopters recognize that keeping the right humans in the loop during the creation, deployment, and ongoing management of GenAI solutions is a must. No magic wand can get the ideal people in the ideal place at the best possible cost.

• Ensure real return
  Nobody “likes” constantly upgrading enterprise platforms, managing data center power, or tracking laptops. This work needs to be done, even though demonstrating real business productivity improvement has traditionally been difficult for CIOs and their teams. GenAI ROI remains uncertain amid complex pricing models with rapidly changing unit prices. Budget holders are, reasonably, uncomfortable with high levels of uncertainty and are wary of locking into another technology that requires endless investment. All but the most exuberant leaders should expect to justify and often constrain risk exposure by limiting scaling without clear, logical, bottom-line returns on investment. This may slow progress, but it’s the right thing to do for the company.
• Manage data to ensure the bot can deliver value
  For years, pundits and consultants have declared, “Data is the new oil/air/gold/soil/bacon” (even bacon…). With the maturation of GenAI, this cliché has advanced to become an essential business truth. GenAIis ravenous for data-the more, the better, and the cleaner, the better. The average enterprise has hundreds of terabytes of data, but much of it is locked up, walled off, sitting on a sales manager’s laptop or a customer’s mobile device. Getting that data into the maw of an LLM safely and securely is one of the few no-regrets decisions a business can make today.

Trust, faith, and computer science

We asked our respondents to rank obstacles to GenAI deployment, and the top four ranked #1 or #2 include an inability to explain and govern the AI “black box,” a lack of skills, and poor data quality tied at #3 with a lack of trust. The honorable mention goes to an uncertain ROI, which nearly half of respondents ranked as #1.

Humans act based on the stories we believe. The core capability of GenAI is to create, abstract, and illustrate stories with content, calculations, and imagery. Business leaders must ask themselves what they believe in.

“With a neural network that’s architected as a probabilistic model, it will attempt to give you the answer it believes you’re looking for, and that may be wrong. Governing a technology like that really requires a different approach. – Cliff Justice, Partner, US Leader, Enterprise Innovation, KPMG”

Whoa. Suddenly, this got serious-and personal.

Let’s be honest. Content from a box we don’t understand called forth by incantations we call “prompts” does not inspire confidence. We’re trusting that the data we feed the bots and the people who manage the signs and messages from those bots can keep us and our companies “safe.”

The fact is, this minute, “safety” hasn’t been proven. Creating output we believe in is really the task at hand for innovators. Questions about data, skilled talent, and value all spring from doubt. Can we believe this? The biggest hurdle for GenAI, looming largest among the obstacles enterprises face, is-paradoxically-faith.

Enterprise leaders love GenAI’s potential for productivity boosts and human enhancement, but they have to balance that with the risks of employing something they struggle to trust. The hard truth is that there is no surefire path to 100% responsible AI. This is the real challenge at hand.

The Bottom Line: Welcome to the era of innovators with the will to act.

As humans, we’ve waited thousands of years for absolute truth to be revealed, and we’re still waiting.

As business leaders, we can’t wait. Life is full of choices and ambiguity. Deciding we can’t move forward without absolute certainty is deciding to stay on the sidelines as the most impactful revolution in centuries passes you by. Like it or not-and believe it or not-we are all in an era for innovators with a will to act.

About this research

Your Generative Enterprise™ playbook for the future is a HFS Research and Ascendion research program based on more than 20 in-depth interviews and a survey of more than 100 C-suite leaders and practitioners with first-hand experience implementing GenAI in organizations.

Watch out for more, and join us on the journey at Ascendion and HFS Research to access all our research findings.