2018 was a year of reckoning for artificial intelligence (AI), proving that it’s here to stay and will soon be all around us. Although the hype was at an all-time high, Deloitte’s State of AI in the Enterprise 2018 report showed that 82% of early adopters of AI saw positive ROI, and 63% adopted machine learning (ML) as a key technology in 2018.
In 2019, we can expect more serious adoption and value creation for businesses with AI. Based on hundreds of conversations with practitioners and thought leaders, and after many AI projects this year, here’s my take on what to expect in 2019.
AI Will Be Omnipresent In Most Human Interactions
AI has proven itself a powerful human capacity augmenter, powering progress and productivity at many levels in the workplace and at home. With a good part of our lives digitally connected or even controlled, AI will become the augmented way to elevate human experiences through the apps we use, the customer support calls we make and the cars we drive. AI will be leveraged by brands to deliver a highly tailored and personalized experience even in non-digital environments.
Early Adopters Emerge As New Leaders; Laggards Fall Behind
AI is a new class of technology where the fast followers may never catch up. The trend of how systems built with machine learning and AI are constantly learning and constantly improving will only continue to grow with time, so starting late may mean never catching up. Harvard Business Review (HBR) published an article on this challenge, indicating that unlike traditional informational technology innovations, organizations will need more time to ramp up and begin using artificial intelligence to deliver value to business processes.
Value Realization Models With AI Will Go Mainstream
One of the key observations in my personal experience in the industry is that more projects that start with a solution to a business problem delivered more value in the enterprise in 2018. This is in stark contrast to most AI projects, where organizations start with data or models and see what they can do with them, experimenting their way into AI. The reason this takes a while is because most folks who are doing machine learning and data science are really not enterprise IT folks who know what it takes to deploy, manage, measure and integrate into the complex enterprise ecosystems. Focusing on the business process and then sourcing the right building blocks to build an enterprise-grade AI will become the winning formula.
Investments In RPA Will Require AI To Deliver On The Promise
One of the technologies that has capitalized on the promise of AI to power productivity is robotic process automation (RPA). Deloitte’s third annual RPA survey shows that even though a large number of enterprises have already started their RPA journeys, only 3% of them have seen any kind of real scale with robotics (over 50 robots in production). We will see RPA vendors trying to develop AI capabilities to address this and partner with AI companies to deliver the promise of RPA at scale with AI to enterprises.
Data Remains The Challenge And The Opportunity With AI
Even though we create 2.5 quintillion bytes of data every day, according to IBM, useful and good data is still uncommon. It’s a gaping hole being exposed by the explosion of machine learning in enterprises over the past few years. While this creates a new class of talent needed to find the needle in the data stack to power your machine learning projects, this part of the problem will continue to persist in 2019, as foundational data mining methods for use of data for ML haven’t evolved yet. While core low-code/no-code machine learning models and tools are becoming more available, enterprises will have to create and curate useful data to power the models. In a weird way, data — not ML — will become a differentiator again for enterprises.
Talent Remains A Barrier To Adoption, But Democratization Is On Its Way
ML technologies require very highly specialized talent around statistics, math, programming and domain/industry expertise. We regularly saw companies fork over huge amounts of money (some even gave over a million dollars) for good ML talent in 2018, but these talents were likely lured and hired away by the top 1% of the organizations. The rest of the enterprises stood little chance in hiring this quality of talent. In 2019, we will see the democratization of talent needed to deliver ML projects, including ready-to-use ML models from leading cloud vendors, auto-ML tools from AWS or GCP that make model selection/deployment easier and developer-focused platforms to build and orchestrate enterprise AI systems.
AI Will Continue To Create New-Collar Jobs
There was a time when job titles such as social media manager and growth marketer didn’t exist. The advent of the internet and then social media has brought about a plethora of opportunities in the last two decades. AI will get there sooner than later, creating an entirely new class of jobs that will involve training and augmenting machines with the human factor. The gig economy will further enable and accelerate this move. I’ve written about this in the past in regard to financial services.
AI Gets Regulated
Facebook and Google have been dominating headlines lately on their use of user data. With the issues of data rights, use of personal data and bias coming to the surface, there is a plethora of issues that will need to be addressed in the digital age. Is personal data covered by property rights, and if so, should it be treated as such? This debate will open up 2019 as the year when progressive and smart governments will start working toward inclusive regulations that will alter the way organizations deploy AI systems to power their businesses. Responsible AI will take its form as a mainstream requirement for anyone seriously looking at AI as a strategic lever in the digital age.