Unlocking The Transformational Value Of AI

The potential for AI to deliver transformative value is almost unlimited. And yet, accessing that value is by no means a given. So how do we crack the code?

As someone who’s been in the business of deploying enterprise-grade AI solutions since the earliest days of AI—from the inside, as a CIO at Verizon, and from the outside, as an advisor to an AI company ASAPP—I know that our job as CIOs is to get transformational value out of transformational technology. And yet as recently as 2020, McKinsey reported that less than 25 percent of companies are “seeing significant bottom-line impact” from AI.

I believe that there are at least three ways we need to shift our thinking if our organizations are going to mine the full transformational potential of AI:

Instead of: Trying to bolt AI on to the way you already work

Try: Designing AI-native processes and architectures

Bolting new technology onto existing ways of working always seems easier. After all, when you’ve invested considerable resources in creating the processes and architectures that you have in place, how could reinvention possibly be cost-effective? But when it comes to truly transformational technology (as opposed to, for example, the kind of incremental improvement we expect from this year’s processor upgrade), reinvention is essential to realizing truly transformational value. If you want to give AI a chance to change the game for your business, you have to let it actually change the game—which means you aren’t just sticking it to the same old game board. Harnessing the full power of AI means baking it into the core of how your business operates.

Example: When we made the transition to developing cloud-native systems, we did things like “lift and shift” our legacy applications to the cloud—and then wondered why we weren’t getting all the benefits we’d been promised, like continuous delivery and auto-scaling. But this was magical thinking. It wasn’t until we realized that we had to rethink our software development lifecycle and our systems architectures that we began to realize the true benefits to our business of cloud-native apps. A similar thing happened when we moved to mobile app development and made the mistake of thinking we could simply display our web applications on mobile phones. We all know how that turned out.

Instead of: Focusing on replacing humans with machines

Try: Discovering ways that machines can help humans do their jobs better

When it comes to AI, companies have historically framed their options as either full automation, on the one hand, or failure to automate, and handoff to a human, on the other. But this framing ignores the reality that technology in general, and AI in particular, has always been about allowing human beings to do superhuman things, with machines comprising a kind of exoskeleton that enhances our capabilities—as individuals and as teams. When we embrace the unique power that exists at this intersection of human-centered and machine-enabled design, we can unlock unprecedented business value. In other words, in the workplace, machines are collaborators, not competitors.

Example: In customer service, enterprises with large call centers have spent decades trying to automate a few more percentage points of their customer service calls—spending billions and having our customers still say that they hate the experience and asking to ‘speak to an agent’. Instead, let’s use powerful machine learning techniques to learn from the very best agents and then help make every customer service agent as good as your best one. The value proposition of making 100 percent of your agents better far exceeds the value of automating a small percentage of the simplest transitions—and it is hard to put a price tag on keeping your customers happy, rather than keeping them at arm’s length. So-called “agent augmentation” is currently a buzzword, but most enterprises have not fully embraced the idea, and most technology providers have not focused on how to do this at scale.

MORE FOR YOU

Lowe’s DIY Digital Transformation Boosts Sales, Productivity And Profit Sharing

Hewlett Packard Enterprise Names New CTO From VMware In Continued Cloud Push

The Digitally Transformed Workplace: Productivity Paradise Or Orwellian Nightmare?

Instead of: Relying on data alone to train our way out of the pitfalls of algorithmic bias

Try: Prioritizing having people at the design table who are traditionally under- or misrepresented by data 

We know that human bias encoded in training data causes our AI systems to reinforce gender-based, racially-based and socioeconomically-based discrimination. What’s more, the lack of diversity in AI impacts which problems get the attention of technology investors, and whether the solutions that are developed are sustainable and ethical in and of themselves. To ensure that we see the full value of AI, we must make it a non-negotiable success metric for our businesses to both hire diverse staff with AI expertise and empower them to influence critical design decisions. As industry leaders, this means we need to lean into solving the so-called “pipeline problem.” We won’t make progress if we simply blame academia for not sending enough diverse AI students into the workforce; instead, we need to collaborate with our colleagues in that sector to understand the barriers to entry that are keeping more women, students of color, and students from low-income backgrounds from pursuing education and careers in AI. 

Example: Comprehensive reports such as the AI Now Institute’s report on Discriminating Systems: Gender, Race, and Power in AI make it clear that we can’t train our way out of the problem. There’s no way around it: hiring diverse staff with AI expertise and then, critically, empowering them to influence critical design decisions, is a non-negotiable. In my last article, I offered guidance on  how to bring “humans in the loop” to combat algorithmic bias in your organization.

AI is a technology with the potential to create transformational value, but to access that value, for both our customers and our stakeholders, we must be willing to transform ourselves. Hiring the same people and following the same established business processes that served us in the past won’t get us to the promised land.

Source: https://www.forbes.com/sites/judithspitz/2021/10/04/unlocking-the-transformational-value-of-ai/?sh=7f73b5966e2a