Machine Learning: Key to Digital Transformation?
Digital transformation and artificial intelligence (AI) techniques such as machine learning have been two of the hottest technology trends in the last couple of years. As it turns out, having them work together can be even better.
According to a recent survey by Infosys, Human Amplification in the Enterprise, 98 percent of respondents who used AI-supported activities to power digital transformation said it had generated additional revenue for their organizations, with 56 percent saying it made more than $10 million.
Specifically, machine learning has had the largest impact, as 75 percent of respondents said it played a significant role in their organizations’ digital transformation, and that the average amount of time managers spent executing day-to-day activities at the end of an AI-supported digital transformation was 84 percent than for managers at organizations beginning such a transformation. And that’s not all:
- 91 percent said AI helped organizations make processes more efficient
- 88 percent said that AI-led automation efforts freed up time for them to focus on more creative activities
- 85 percent said AI implementations were the biggest driver of productivity increases in their organizations
- 84 percent said AI simplifies their work
This is also supported by predictions from analysts such as IDC and Gartner. For example, by 2019, 40 percent of all digital transformation initiatives will be supported by cognitive/AI capabilities, and at least 20 percent of all workers will use automated assistance technologies by next year, predicts IDC.
Similarly, Gartner has predicted that by next year, 20 percent of business content (such as shareholder reports, legal documents, and press releases) will be written by machines. “Technologies with the ability to proactively assemble and deliver information through automated composition engines are fostering a movement from human- to machine-generated business content,” the company writes. “Data-based and analytical information can be turned into natural language writing using these emerging tools.” That will enable digital transformation because humans will need to be involved less in those business processes.
In addition, by 2020, autonomous software agents outside of human control will participate in five percent of all economic transactions, the company predicts. “Algorithmically driven agents are already participating in our economy. However, while these agents are automated, they are not fully autonomous, because they are directly tethered to a robust collection of mechanisms controlled by humans — in the domains of our corporate, legal, economic and fiduciary systems,” the company explains. “New autonomous software agents will hold value themselves,” particularly in the banking, financial, insurance, markets, exchanges, and crowdfunding industries. Recent forecasts show the AI market for the financial services industry specifically is expected to grow at 40.4 percent CAGR in the next half decade, with a proposed $7.306 billion market value in 2022, writes Jason Koestenblatt for the Enterprise Mobility Exchange.
Even the Central Intelligence Agency is getting into the machine learning/digital transformation act. “AI and machine learning won’t replace the human element of intelligence — as might be the case with personnel in other sectors — but instead it will boost agency employees’ performance by eliminating some of the more minute but labor-intensive aspects of their mission,” writes Billy Mitchell in FedScoop.
Machine learning has been used to analyze much more data than a human could, coming to more, and better, conclusions faster. Increasingly, companies are using those conclusions to digitize business processes, not only making decisions faster but also acting on those decisions without a human having to be involved, streamlining the business process.
“Logistics companies, online retailers and fulfillment centers on the digital frontier already rely on AI-powered robots to run their operations,” writes Orlando Trott in Medium. “In some cases, organizations have integrated their systems to auto-order inventory when product is low. Agriculture is testing AI in self-driving or program assisted tractors while consumer packaged good marketing departments are using AI to detect and direct advertisements towards very specific audience groups based on location and personality type.” When integrated correctly and paired with AI, businesses can experience streamlined workflows and become more efficient, he adds.
Even tasks within IT itself can be automated using machine learning, writes Alison DeNisco in ZDNet. “Tasks such as backups, job scheduling, password resets are all in the process of being automated,” she writes.
Typically, machine learning has been based on humans telling the machines what to do, but at this point, machines are getting good enough at it that they are figuring out what conclusions they can make from data on their own, writes Evan Schuman in Ars Technica. “Companies are looking to machine learning to automate processes or to augment humans by assisting them in data-driven tasks,” he writes. “With minimal pre-task human efforts needed, the scalability of unsupervised machine learning is much higher.”
“Unsupervised” doesn’t mean you can just throw the machine learning system at your company’s entire corpus of data and have it come back with the secret to the universe, Schuman warns. A data scientist still needs to choose which data the system should look at, and then analyze the groupings of data that the system finds, he writes.
At the same time, it is important for humans to continue to supervise machine learning to ensure that the machine doesn’t make decisions that are unethical or discriminatory, Schuman writes. For example, while a system could find a correlation between consumers defaulting on their loans and those who have blue eyes, it would be discriminatory if the software decided to scan every customer’s eye color and use that information to decide whether to approve a loan, he writes.
For that reason, diversity in business – such as data, talents, backgrounds, and cultures — is more important than ever, writes Divina Paredes in CIO. “Diversity of viewpoints counteracts the ‘dark side’ of AI, which is bias,” she writes. “Diversity will allow us to overcome this bias and to harness the power of the crowd.”
