3 Realities of AI, and What 50+ Execs Say About Them
Monday, January 14, 2019
Posted by: Kaylin Berg
Ryan Maguire, CTO of Emerging Technologies - AI, Solstice
In November at Solstice’s global headquarters in Chicago, we brought together cross-industry thought leaders to have a discussion about artificial intelligence (AI), the disruption it has already created, the impact it will have next, and how businesses can start their AI journey. The conversation touched on many of the most pressing challenges surrounding AI: technology, business problems, and outcomes, demystifying the hype, data maturity, and overcoming obstacles along the way.
With the participation of over 50 C-Suite executives, a cross-vertical panel of experts led our discussion, including Adam Kanouse, CTO at Narrative Science; Greg Mikels, customer engineering and machine-learning specialist at Google; John Pettit, CTO at Backstop Solutions Group; and our Principal Digital Strategist Jared Johnson.
The Right Problems for AI
The first - and seemingly most prevalent - obstacle that business leaders face today is identifying the right problems that AI can solve. Many attendees shared feelings of paralysis, knowing that AI may have the potential to drive change but struggled to identify how to do so or how to secure the investment required. Executive alignment and support is never an easy task, but in the often immeasurable world of AI, leaders must confront a heightened challenge to gain that buy-in and investment.
While this challenge is formidable, our panelists shared a few tips to move forward and win over stakeholders. First, focus on outcomes, not the technology. Start by building a scorecard to evaluate opportunities and grade them based on the maturity of your business, your industry, and the technology required. Additionally, don’t wait for the perfect solution with the perfect data to fall onto your plate. As with all emerging technology, the key is to incrementally build value and continue to learn and grow along the way. In going through this exercise with clients before, we’ve learned a lot and have used those lessons to build our Applied AI Strategy engagement.
Blocked by Data Maturity
Another obstacle was not feeling data-mature enough to begin leveraging AI technologies that are highly data dependent. For many, AI feels unachievable because data is sparse, low-quality, and disparate. The discussion proved this problem is far-reaching—and despite access to more data sources than ever before, including IoT, user analytics, and social data, the problem isn’t going away. Our panelists acknowledged that enterprises need to address this gap now. However, enterprises must first acknowledge the problem and start making investments to level-up and become AI ready. Approaching this challenge while keeping the outcome-oriented and incremental growth mindset, shared earlier, is the only way to learn more about the creation of an AI-worthy data environment.
Uncertainty and compliance in an AI world
Finally, as companies start to invest in AI, identify outcomes, and improve their AI readiness, attendees shared uncertainty around compliance and regulation. Attendees wondered how, as machine-learning-model accuracy continues to improve, we could ensure performance would continue. And they wondered how, even if the performance and accuracy remain high on paper, we can be sure the model is making the predictions that people and businesses expect. If models can’t be trusted for business-critical decisions or fail to live up to the expectations of humanity, then what? Our panelists acknowledged that this is a point of struggle for many businesses that have begun AI transformations but, again, offered advice on how to remain compliant and ensure satisfaction.
Like other technologies, AI isn’t “set it and forget it.” Mature enterprises need to install guardrails and constantly measure performance and adjust expectations. Companies such as Uber, Facebook, and Google have begun building continuous evaluation frameworks. At Solstice, we deploy our process called “Continuous Learning.” Continuous Learning is a framework that leverages CXDD, modern architecture and retaining humans in the loop to continuously evolve models, measure accuracy, and assess the risk to ensure that all expectations will continue to be met.
While attendees had an opportunity to ask some of the questions facing their AI transformation, we know that many enterprises have reached a dead end. Getting stuck in the AI maze and struggling to make it to the end is a feeling shared across industries and across digital maturity. We know that AI will enable the future of digital solutions. Our mission is to find the path to value, working side by side with our partners to get out of the AI maze and to Make it Real.
To learn more about our AI practice, visit solstice.com/ai, or reach out to me directly at firstname.lastname@example.org.
This post originally appeared on the Solstice blog, here.