It's a Wednesday evening, you're winding down, or at least you think you are. The last few steps before bed include asking Alexa to reorder toilet paper from Amazon, checking Facebook for the latest posts, and discovering a new TV series to binge-watch on Netflix.
We experience the power of artificial intelligence every day as consumers, but do you take advantage of AI or data science in your day-to-day business? Probably not.
IDC forecasts that spending on AI technologies will grow to US$97.9 billion in 2023—more than two and a half times the spending level of 2019. In addition, a Deloitte study reflects 45 percent of seasoned AI businesses believe AI and machine learning technology enables them to have a significant lead over their competitors. This dynamic suggests that the rich are getting richer as it relates to applying the latest AI capabilities and machine learning technology into their businesses, and gaining a competitive edge in their respective markets.
Pretty soon, AI will become the way companies win.
The good news is that innovation is already well on its way with the number of AI applications increasing every day - autonomous vehicles, chatbots, healthcare applications like increasing clinical trial efficiency, customer experience applications like product recommendations to name a few. For simplicity, we categorize the way businesses can use AI into three categories: power products and services, automate operations, and improve decision making.
This is the most common category. Companies whose core value proposition and business model hinges on the value that AI or machine learning brings to their solution. Think Google search, or the algorithms powering Amazon's logistics network. This isn't reserved for the tech titans though. Many SaaS or fintech companies have also created successful businesses intentionally designed around a specific AI application. Most on-staff machine learning engineers and data scientists tackle problems like this - enabling product development through AI.
This is an emerging category. The objective is to take high volumes of data and automate complex business processes that typically require manual review or data entry. By doing so, companies can scale their operations much faster, and draw more insight by real-time pattern matching across Big Data rather than static ad-hoc analyses. An intuitive application is automation for a manufacturing floor where supply chains use computer vision to detect failures during inspection processes. Other examples that apply to almost any business include prioritizing the best marketing leads or routing customer tickets for efficient resolution. It's a naturally emerging category following revenue-generating product development use cases, but one that is imperative with the pandemic highlighting how important speed and agility are to sustain a business.
This is the most misunderstood category. Data informs insights, insights drive decisions, and decisions drive trajectories for businesses so it's no question the value is immense. Harvard Business Review discusses the importance of Big Data, AI, and machine learning-driven decision-making in staying competitive. However, all too often companies currently rely on reporting, business intelligence, or ad-hoc data analytics using spreadsheets to draw insight based on historical data available at a point in time. The challenge is that these analyses become stale and are resource-intensive if done well, meaning a few projects get the most attention. Ideally, an AI business leverages machine learning to create living models that forecast the future while learning the change in your business and globally. For example, it's no good to analyze the factors of historical customer churn without a way to predict and address an at-risk client before they leave.
Now that we know how AI can be used to drive growth for businesses, how can you start applying it to yours? Here are 4 easy steps to follow.
There's a lot of AI hype. Just search for it on Google and you'll find tons of results spanning every industry. The reality is, not every business is a perfect fit. If you're in experimentation mode as a startup or as a project leader in a larger enterprise, you likely don't have enough historical outcomes in your datasets to apply machine learning. In addition to data availability, data-driven cultures will have the most success. The end game is to action behavioral change based on the output of AI so organizations must inherently embrace data and technology. If your organization isn’t there yet, that’s step 1.
So how do you know if AI is right for you? Here are three key questions to ask:
1. Do you have enough data?
2. Is your culture ready for data-driven decision making?
3. Do you have a use case that can benefit from AI?
If the answer to all three questions is "yes," then you're ready to start exploring AI solutions. If not, don't force it. The last thing you want is to be part of the 85% of AI projects that fail.
The key limiter is a talent shortage in data scientists and machine learning engineers with real-world experience. Deloitte mentions that 23% of seasoned AI adopters report an extreme or major talent gap. Even Amazon is now repurposing internal talent to try and close the gap. So, be prepared to invest in a very competitive talent pool. If your core IP is built around AI or you're a large enterprise with a budget to put behind it, then it may be worth it.
Weigh the pros and cons of your AI initiative and decide if you have the bandwidth to take it on internally. Consider factors such as:
- Are you able to build and maintain an in-house data science team?
- Do you have the compute resources available to train and deploy models?
- Do you have a long enough timeline to accomodate creating the infrastructure and expertise necessary for success?
If you answered yes to most of the above, then building in-house may be your best bet. However, be warned that this is not a decision to take lightly. Building an in-house AI organization is no small feat.
Your "top 3" projects already have plenty of firepower behind it with dedicated engineering resources and/or consultants on staff. Starting with a tier 1 project also isn't an ideal way to introduce AI for the first time. The next slug of projects that didn't meet the cut are great candidates. They have meaningful business value, typically get sidelined or forgotten about, and are perfect test beds without the Tier 1 project spotlight to prove out the value AI can bring to the business. Perhaps you already use AI to power your products and services, but don't yet use AI for your enterprise business functions (e.g. finance, sales, marketing, customer success, operations). That's also a great place to dive in.
Projects that didn't make the cut can be a great way to test out AI without the pressure of a Tier 1 project.
There can be many reasons why a project gets sidelined or forgotten about. Maybe it's been on the backburner for too long, or maybe it just doesn't have the resources behind it to get it off the ground. Whatever the reason, these projects can be great candidates for AI.
Diving into AI with a less critical project can help you learn about the capabilities and limits of AI.
There's no shortage of tools out on the market for you to start using AI but the level of technical expertise and coding experience varies greatly. For projects that can afford large data science or machine learning engineering teams, platforms like DataRobot, C3, or H2O will provide the most robust tooling and even the tech titans (Amazon, Microsoft, and Google) all have their own AutoML tools. For quick proof of concepts or projects that can't afford those resources, no/low code tools like Akkio, Obviously.ai, and Levity are a great fit. Fast and easy to use and feel more like spreadsheets, which analysts or business end-users are more accustomed to.
Once you've found the right tool, it's important to get buy-in from your team. AI is still a very new and foreign concept to some business users and there can be a lot of fear around implementing it. Be sure to sell the benefits of AI and how it can help the team be more efficient or improve decision making. Most importantly, continue to communicate and iterate on the project as it progresses. The beauty of AI is that it's always learning and evolving, so there's always room for improvement.
IBM Watson was introduced in 2010 and started the wave in personifying AI but in many ways, has still left the majority of business leaders perplexed at AI and machine learning. Don't be intimidated by the mystical language of neural networks, natural language processing, or deep learning. There was a point in time when spreadsheets were just as foreign too - but what business doesn't use spreadsheets now? AI systems will become just as common before you know it. Business leaders and analysts - start your “AI strategy” by leveraging AI for one bite-sized project. Get your long-term competitive edge today and have some fun while you’re at it.