Major players in the tech industry are pushing the boundaries of self-determining computers, especially as cutting-edge technologies like artificial intelligence (AI) and machine learning become more mainstream. While many professionals understand that these technologies will make their jobs easier – or even take over specific tasks – others are feeling confusion. One common question is, what’s the difference between machine learning and automation?
Let’s start with machine learning, a subset of AI.
“It’s an evolution,” said Andreas Roell, managing partner of Analytics Ventures, a consultancy that helps businesses adopt AI. “AI fits into the bucket of workload analysis or task analysis. Business intelligence also sits in that same bucket. It’s taking data, then analyzing it.” [Related article: How Artificial Intelligence Is Transforming Business]
Machine learning is typically a later-stage development, where machines take in data on their own and then analyze it, Roell explained. The biggest difference is that “machine learning identifies data signals relevant for the future,” he added.
Automation is frequently confused with AI. Like automation, AI is designed to streamline tasks and speed workflows. But the difference is that automation is fixed solely on repetitive, instructive tasks, and after it performs a job, it thinks no further.
There’s a good chance you use automation without even realizing it – for example, by automating emails to customers, automating the way you generate invoices, or automatically logging a help-desk inquiry. Workplace automation saves time and allows workers to focus on higher-priority initiatives. It’s a reliable, computerized workhorse, able to show up and get the job done.
Machine learning takes these tasks and layers them in an element of prediction. Whereas automation would continue to do exactly as you requested – say, send invoices on a specific day – machine learning predicts when the invoices should go out, who did or did not receive one, and when payments are on the verge of being late.
No, AI and automation are not the same. Automation involves an entire category of technologies that provide activity or work without human involvement. For example, say an old-style water wheel represents automation, translating the power of falling water into a repetitive nonhuman activity or mechanical work. There is nothing about the water wheel that involves artificial intelligence; it just keeps doing the same thing over and over.
We often associate automation with computers, but it’s been around for ages.
“If you can take the resources that you have and come up with some sort of silver bullet and that turns them into radically better efficiency for what you’re getting back, that is going to be evolutionary dynamite,” zoologist Antone Martinho-Truswell told Gizmodo. “You’re going to do fantastically well, as we have. Our nearest relatives are all endangered because of us.”
AI, on the other hand, involves a machine exhibiting and practicing something similar to what we describe as human thinking – that is, the ability to interact in thousands of ways with the world around us without receiving any prior explicit coding or instructions. Think, for instance, of how AI digital assistants like Siri or Alexa can understand and respond to our questions and commands.
The rate at which companies adopt AI is continuing to grow. Companies that have adopted AI are finding extensive cost savings, according to the McKinsey report on the state of AI in 2021, which interviewed 1,800 business leaders from various industries worldwide.
Machine learning works to understand data, leveraging what Roell called data signals to drive future intelligence. It’s not simply performing an “if X, then Y” task stream; it’s essentially thinking through data much like a human.
“There’s a lot of fear around AI, that it will eliminate jobs,” Roell said. “That’s not what it’s supposed to do; it’s making the way we work easier. But what it will do is lead to entirely new categories of jobs being created.”
Roell gave the example of call center employees now being used to categorize the vast amounts of data used by AI. Several companies have taken this approach.
“Now that is true innovation,” Roell said.
Machine learning can be automated when it involves the same activity again and again. However, the fundamental nature of machine learning deals with the opposite: variable conditions. In this regard, machine learning must be able to function independently and with different solutions to match different demands. There is a higher likelihood that machine learning would be applied to determining unknown prediction scenarios.
However, the principle could apply in automated systems as a safeguard or as an element of automation, according to the Brookings Institution. For example, a computer system used to move Amazon.com boxes could learn millions upon millions of weights so that it could flag a box on the conveyor belt that doesn’t match known inventory when it senses the anomaly along the way from the shelf to the shipping truck.
Not quite, according to Stanford University’s Artificial Intelligence Index Report 2022. AI systems are a business tech trend being more broadly deployed into the global economy. With this increasing deployment comes a commensurate increase in AI capabilities. Language models are becoming ever more accurate, image classification training times are quickly decreasing, and AI systems are increasingly affordable.
However, there are still substantial gaps between AI systems focused on particular activities and general-purpose thinking AI systems.
Adam Uzialko and Joanna Furlong contributed to the writing and reporting in this article. Source interviews were conducted for a previous version of this article.