The AI aspiration gap: How to hook AI into actual business results | Industry
Getting your AI deployment right takes five steps: assessing your data strategy, aligning stakeholders, assessing tech feasibility, and a coordinated approach to ethics. To learn more about the five fundamentals of AI readiness, and ensure you get it right at every step, catch up on this VB Live event.
Access on demand for free right here.
AI is a roughly $8 billion market today, but expected to exceed $106 billion within just a couple of years, says Jessica Groopman, industry analyst and founding partner of Kaleido Insights. And with so many products on the market, it’s incredibly difficult to separate the wheat from the chaff, and make sense of what’s worth pursuing and what’s just another bright, shiny hyped-up object rolling by.
AI is driving commercial applications and igniting tremendous change around traditional customer experience, as well change in how marketers, sales agents, and business analysts are using data. It’s disrupting the way consumers interact with brands and how brands can find consumers where and when they want to be contacted. Automation and AI impact every single phase of the customer journey, from the introduction of visual search and discovery in the research phase to new and impactful customer interfaces and transactions designed around experiences like biometric payment and facial recognition.
“There’s a lot of buzz and hype in this space right now, but when you look under the hood, very few businesses are doing this at scale or across multiple business functions,” Groopman says. “There’s this big gap between the aspirations and what’s happening on the ground.”
While about 80 percent of businesses are running AI in some form, in some area of the business and production, a third of projects are actually succeeding — many just pilot proof of concepts that don’t necessarily see the light of day. Another recent survey from MIT Sloan found that less than 30 percent of companies have any kind of AI strategy, which may have something to do with that low success rate.
“But achieving AI has to require more than just understanding specific use cases,” Groopman says. There’s no necessary starting point. Companies are starting all over the map. Very often AI emerges from within a single business function. Sometimes it’s an innovation team that has a charter to go apply or test out some new capabilities using open-source tools. Sometimes it’s a rogue business analyst who just wants to try something, or has maybe been allocated five or 10 percent experimentation time. Sometimes it’s a formal program that says, here’s where we have the largest, cleanest data set. Let’s try out some process efficiencies. But very often it starts from the bottom rather from the top, a ground-up approach.
“That said, it’s important that companies do have a strategy and think about AI as not just another pillar or technology component to add on, but as part of a much broader data strategy,” she explains. “AI is not a bolt-on. It’s not a different silo. It’s fundamentally about doing more with data and trying new software development mechanisms, new capabilities, new predictive capabilities rather than just historic data analysis.”
While the strategy will look very different for every company, it also needs to be about laying the foundations for governance, as well as the foundations for processes to make the most of these investments, whether in data or applications or talent.
Companies tend to be hyperfocused on preparing data and prepping data pipelines, with an eye on hiring data science talent at some vague point in the uncharted future, Groopman says, but if you look back on the impact that these technologies have had on businesses and process or product transformation, you can see it’s much bigger than a data story — it’s a culture story.
Data is a requirement, a prerequisite, but it’s about selling people on AI, from leadership to the salespeople who are supposed to leverage these new insights or new information. Frontline associates, customer service agents, field technicians — anyone on the front lines of the application. LIke subject matter experts, whether that means a doctor in health care or a cybersecurity expert, who are especially essential once you begin to automate decisions into algorithms. And end users, whose experiences need to be baked into the training of machine learning models. And designers, especially from a UX standpoint, but also from a systems and application design standpoint. And of course, product leaders, the essential liaisons between communicating the core needs of a product back to leadership.
“One of the things that we surfaced was how to ready each of these groups,” says Groopman. “They’re different. They have different communication needs and priorities, and there are best practices for how to really prepare each of these different personas. Preparing frontline associates looks very different from preparing leadership, as you might imagine.”
Beginning to add AI to your organization is relatively easy, Groopman says. There are a variety of tools, via both vendors and open source, that are available to begin to pilot specific challenges, specific problems, and learn from those findings. But in the same way that AI should not be just thrown in the mix, AI preparedness cannot be just a bolt-on. Rather it must be seen as an extension or evolution of what your organization is doing with data in the first place, and then looked at in the broader story.
“Machine learning should translate to enterprise learning over time, or company learning for that matter,” Groopman says. “It’s not just a technical build. Investment in AI is investment in people. People are an essential part of making these tools better and deploying them and making sense of them and defining value and assigning metrics. We are just as inherent as the algorithms are.”
For a deep dive into real-world case studies from $100-million retailer evo and the company behind the woman’s megabrand Natori, along with pragmatic and practical advice from the experts who have gotten their hands dirty, as well as the top five best practices for AI implementation, catch up on this VB Live event!
Don’t miss out!
Access on-demand right here.
Watch this webinar and learn:
- What you need to do to prepare for AI– beyond the data science team
- Real-world examples and research findings
- Top 5 best practices for strategic AI implementation
- Nathan Decker, Director of eCommerce, evo
- Ken Natori, President, Natori Company
- Jessica Groopman, Industry analyst and founding partner of Kaleido Insights
- Rachael Brownell, Moderator, VentureBeat