Your Top Artificial Intelligence Adoption Questions, Answered

According to Gartner, the number of organizations implementing some type of artificial intelligence (i.e. machine learning, deep learning and automation) has grown by 270% over the past four years. One big reason for this boost is the fact that executives and decision makers are beginning to recognize the value that these innovative technologies present.

That’s not to say they’re all on board. Are CEOs getting savvier about AI? Yes. Do they still have questions? Also yes – particularly as it relates to the adoption/deployment process. Let’s take a look at a few of the top questions and answers surrounding the topic of artificial intelligence below, along with some practical advice for getting started.

Is a business case necessary for AI?

Most AI projects are viewed as a success when they further an overarching, predefined goal, when they support the existing culture, when they produce something that the competition hasn’t and when they are rolled out in increments. At the end of the day, it’s really all about perspective. For some, AI is viewed as disruptive and innovative. For others, it might represent the culmination of previous efforts that have laid a foundation.

To answer this question, examine other strategic projects within the company. Did they require business cases? If so, determine whether your AI initiative should follow suit or whether it should be standalone. Likewise, if business cases are typically necessary in order to justify capital expenditure, one may be necessary for AI. Ultimately, you should determine exactly what will happen in the absence of a business case. Will there be a delay in funding? Will there be certain sacrifices?

Should we adopt an external solution or should we code from scratch?

For some companies, artificial intelligence adoption came at the hands of dedicated developers and engineers tirelessly writing custom code. These days, such an effort isn’t really necessary. The problem is, many executives romanticize the process, conveniently forgetting that working from scratch also involves other time-intensive activities, like market research, development planning, data knowledge and training (just to name a few). All of these things can actually delay AI delivery.

Utilizing a pre-packaged solution, on the other hand, can shave weeks or even months off the development timeline, accelerating productivity and boosting time-to-value. To determine which option is right for your organization, start by defining budget and success metrics. You should also carefully assess the current skill level of your IT staff. If human resources are scarce or if time is of the essence, opting for a ready-made solution probably makes the most sense (as it does in most cases).

What kind of reporting structure are we looking at for the AI team?

Another thing that’s always top-of-mind with executives is organizational issues, specifically as they relate to driving growth and maximizing efficiencies. But while this question may not be new, the answer just might be. Some managers may advocate for a formal data science team while others may expect AI to fall under the umbrella of the existing data center-of-excellence (COE).

The truth is, the positioning of AI will ultimately depend on current practices as well as overarching needs and goals. For example, one company might designate a small group of customer service agents to spearhead a chatbot project while another organization might consider AI more of an enterprise service and, as such, designate machine learning developers and statisticians into a separate team that reports directly to the CIO. It all comes down to what works for your business.

To determine the answer to this question, first figure out how competitively differentiating the expected outcome should be. In other words, if the AI effort is viewed as strategic, it might make sense to form a team of developers and subject matter experts with its own headcount and budget. On a lesser scale, siphoning resources from existing teams and projects might suffice. You should also ask what internal skills are currently available and whether it would be wiser to hire externally.

Practical advice for organizations just getting started with AI:

Being successful with AI requires a bit of a balancing act. On one hand, if you are new to artificial intelligence, you want to be cautious about deviating from the status quo. On the other hand, positioning the technology as evolutionary and disruptive (which it certainly is) can be a true game-changer.

In either case, the most critical measures for AI success include setting appropriate and accurate expectations, communicating them continuously and addressing questions and concerns with swiftness and transparency.

A few more considerations:

  • Develop a high-level delivery schedule and do your best to adhere to it.
  • Execution matters, so be sure you’re actually building something and be clear about your plan of delivery.
  • Help others envision the benefits. Does AI promise significant cost reductions? Competitive advantage? Greater brand awareness? Figure out those hot buttons and press them. Often.
  • Explain enough to illustrate in the goal. Avoid vagueness and ambiguity.

Today’s organizations are getting serious about AI in a way we’ve never seen before. The better your team of decision makers understands about how and why it will be rolled out and leveraged, the better your chances of successfully delivering on that value, both now and in the future.