In many ways, we’re close to a tipping point in the development of artificial intelligence technologies, with fleets of self-driving cars, hive drones, automated retail experiences, and more. Companies of every size now must grapple with how AI will affect and possibly eliminate their sector, and how they can adapt and leverage AI and machine learning (ML) to disrupt themselves before they get disrupted.
While it’s crucial for companies to start participating in the AI movement or risk getting left behind, it’s important to do so strategically to ensure AI initiatives are truly helping to achieve ultimate business goals. AI-ML will make many of our day-to-day lives better and more productive by augmenting what we do already in much more accurate and efficient ways.
Tech innovators like Uber, Google, and Facebook are making big bets on AI through acquisitions and internal organizational makeovers that make AI a strategic priority. At Davos 2017, world leaders debated the technology’s looming impact on jobs and which industries are ready to adapt.
Indeed, corporate spending on AI technology will surge to $47 billion in 2020, from $8 billion last year, according to IDC. By 2019, over 110 million consumer devices with embedded intelligent assistants will be installed in U.S. households, IDC projects.
Intuit’s AI transformation began six years ago when we asked ourselves, “What is the future of financial software, and what role do we play helping customers?” The trend was clear even then — personalization, automation, augmentation, and speed, all of which required software that does the work with and for users. As AI advances came together over recent years, we refined our vision.
We view AI as an opportunity to eliminate drudgery for our users. We expect AI to help power a multiyear effort to create technology so that taxes are “done” for consumers and accounting is “done” for small businesses in ways no one could imagine a few years ago. But getting there is no small task. Intuit has more than 30 AI and ML models in production, touching customers in real-time experiences, and we’re still experimenting with what is possible. We’re constantly learning, evolving, and improving based on our experiments and outcomes.
As your own company works to build products and services that answer real consumer needs with AI, here are several key things to consider.
1. Get serious about your data
Every company’s AI journey must begin with data. Data — rivers, lakes, and oceans of it — is what powers machine learning and artificial intelligence. If you’ve been thinking about how to leverage your data, consider it now urgent to develop and execute on a plan.
At Intuit, we began transforming our approach to data years ago. We set a roadmap after talking to our business unit partners to understand our customers’ needs, industry trends, and the key customer benefits that could be enabled through data. Our centralized Intuit Analytics Cloud broke down the many silos that enabled us to look at the reams of data our products generate in a more holistic way. We started before we had any AI or ML models in production, focusing on foundational work around insights and personalization. Reliable, fast, and accurate data is truly an accelerant to AI-ML outcomes for our customers.
2. Look into your own crystal ball
It’s important to envision your own future as a company and how AI could empower you or disrupt your current thinking about development. At Intuit, we went through a visioning exercise in which we met with a wide variety of organizations and individuals — from university experts to startups, VC firms, and more — to get a well-rounded perspective on the AI and ML landscape and how Intuit fits in. Smaller organizations can do the same at a scale that is appropriate to the company’s resources.
3. Instill the right mindset
It can be challenging early on to explain the benefits of investing in AI to business unit and product leaders. It’s important to demonstrate value as customer benefits every quarter or so in a business context, rather than a data scientist context. For example, Intuit’s Tax Knowledge Engine analyzes the tax returns of 30 million customers to make a recommendation on itemizing versus taking the standard deduction, which can save as much as 40 percent of tax prep time. That’s a tangible business outcome that a non-data scientist can understand. AI-ML is a journey. It’s important to set the expectations of potential outcomes while verifying via real and tangible impact.
4. Draft a team and pick a project
To get started on AI in your enterprise, pick a low-hanging business problem, form a cross-functional team including data scientists, engineers, and QA associates, and iterate fast. Build a culture that understands the value of experimentation. Cherish the learnings you achieve by applying AI to your offerings. Your team should consist of people with domain expertise and real world experience. Don’t feel like you need to hire only deep learning experts. Depending on the problem you want to solve, balance them with other types of scientists — people with PhDs in other areas who are inherently curious, understand the principles of science, and have different approaches to resolving problems. Real AI-ML can deliver significant outcomes today with applied data science. Don’t underestimate the importance of that significance while driving the mindset change.
We’re only at the beginning of what’s possible with AI and machine learning. Establishing best practices at the start of your company’s journey will help put you in a position of strength for what’s ahead.