Published: Jan 16, 2020
By Rob Preston
Truly "autonomous" systems are starting to replace or augment many of the routine tasks and processes people perform every day, improving efficiency while freeing individuals for higher-level pursuits.
We're all familiar with the highest-profile example: self-driving cars. But what's often overlooked is how much progress is happening in other areas and industries: healthcare, air travel, energy provision, retail, logistics, agriculture, and construction. Autonomous systems are even helping governments match refugees with the most suitable communities to live, as detailed in one of the four real-world vignettes we present below.
When we asked in our recent survey of 518 people in a variety of industries where they expect "significant" use of autonomous technology within five years, six sectors rose to the top:
Such optimism makes sense, given advances such as self-managing and self-patching databases in IT.
But our survey's other findings might underestimate the pace of change: Just 24% say they expect to see significant use of autonomous tech in construction, for example, even though self-driving bulldozers already are in use on select projects.
Today's autonomous systems go beyond mere automation. They use machine learning, a branch of artificial intelligence, to analyze conditions and continually learn from the massive amounts of data they ingest, enabling them to respond—often in real time—to all manner of questions and changing circumstances in order to make decisions and take actions on their own.
Autonomous systems aren't merely the future; they've arrived. What follows are four of the most compelling examples, with complementary insights from our extensive survey.
Once it's finished, the site will be ready for workers to start building the foundations for wind turbines that will stand more than 200 feet high.
This isn't the future. It's happening today on select wind and solar power projects that Mortenson, a US construction, development, and engineering services company, manages in remote areas. Mortenson operates the autonomous equipment in partnership with Built Robotics, a startup whose on-board technology turns a conventional excavator or bulldozer into a self-operating robot.
Unlike self-driving cars, where a vehicle has to learn every possible surprise that might arise on the road, an excavator doing one thing well, working in a geo-fenced area, faces far fewer obstacles. It's why controlled environments such as a construction site will be among the first places we're likely to see autonomous systems gain widespread adoption.
”With self-driving cars, you can't just drive straight on a highway. You need to know how to do everything for it to be useful," says Gaurav Kikani, vice president at Built Robotics. "We can take an application like trenching or excavation, and if we just automate that, we can start creating value."
Most people aren't aware of this potential. Asked how soon we'll routinely see self-guided equipment doing common jobs such as digging and grading at a work site, 60% of the respondents to our survey say they think it will be five years or longer.
Three big challenges set the construction industry's innovation agenda: improving safety, combating a labor shortage, and increasing productivity, says Burcin Kaplanoglu, executive director, innovation officer, of Oracle's Construction and Engineering unit.
For leaders like Mortenson, the autonomous opportunity is clear. "Our goal is to embrace the change that is happening in our industry—to help create safer working environments for our team members and value for our customers," says Eric Sellman, vice president and general manager of the company's Civil Group.
In terms of safety, if autonomous machines can take over certain jobs, it could be a big selling point. In fact, reducing errors and accidents is considered one of the most important benefits of autonomous technology, cited by 53% of survey respondents. But companies will hold autonomous robots to a higher safety standard than they hold human operators, Kaplanoglu says.
Meanwhile, because construction companies struggle to find enough skilled workers, heavy equipment often sits idle because there's no one to run it. "Through autonomy, we can unlock the full 24 hours in a day in terms of equipment utilization," Built Robotics' Kikani says.
Which kinds of construction tasks will autonomous equipment do? Likely "dull, dirty, and dangerous work," says Kaustubh Pandya, a principal with Brick & Mortar Ventures, an industry venture capital firm. Technical feasibility in autonomous vehicles has rapidly advanced, "though we need to shape it to meet the constraints of construction," Pandya says. "We have to be able to retrofit old equipment, operate in low connectivity areas, and mix autonomous and semi-autonomous equipment."
Balancing all these factors is a management challenge unto itself. Says Kaplanoglu: "In many cases the technology is moving faster than the people and the processes."
—Chris Murphy
If doctors catch it early, before symptoms appear, they have a good chance of saving a patient’s eyesight. But many diabetics never get around to visiting a specialized eye doctor who can spot the warning signs.
Most diabetics do visit their primary care physician regularly. What if there was a way to test for retinopathy at those checkups?
Enter IDx-DR (pictured above)—a device that, with a simple eye scan, can diagnose whether a person has retinopathy, without the judgment of a doctor. It’s the first artificial intelligence-based diagnostic device approved by the US Food and Drug Administration.
For Dr. Michael Abramoff, the retinologist and tech entrepreneur who created the device, IDx-DR exemplifies the potential for autonomous healthcare systems. “Autonomous systems will increase productivity, which will drive prices down and provide better access to healthcare—some parts of the United States don’t have a lot of doctors,” he says. “These great benefits come because you can make these systems available anywhere.”
Diagnoses are only one way autonomous systems could change healthcare. Such systems could speed up the allocation of equipment, beds, doctors, nurses, and orderlies across hospitals without human assistance.
Where are autonomous healthcare systems most likely to appear next? For starters, one good candidate is analyzing images or scans, even if a doctor continues to make the final decision on a treatment or diagnosis. “Hopefully, in a few years we’ll see an autonomous system that can even prescribe medications,” Dr. Abramoff says.
In our survey, 39% of respondents say they think we’ll routinely see technology that makes a diagnosis without a doctor in the next four years or sooner.
Healthcare providers will understandably be cautious about implementing autonomous systems, and they will need to address patients’ concerns about safety and employees’ worries about job security. But Rebecca Laborde, lead product strategist for healthcare and precision medicine at Oracle, recently talked with a hospital executive who had lots of ideas for redeploying staff if they could spend less time on repetitive tasks.
For example, hospital administrators “could participate in clinical trials that they haven't had the bandwidth to take on,” Laborde says. “They could investigate new technologies, such as trying out chatbots on the hospital website. They could get more involved in community health initiatives in their area.”
Accenture principals Paul R. Daugherty and H. James Wilson, in their recent book Human + Machine: Reimagining Work in the Age of AI, take issue with the widespread belief that workers will necessarily compete with increasingly intelligent, autonomous machines. To the contrary, their research finds that artificial intelligence complements and augments human capabilities—AI-powered machines do what they do best (transact, predict, iterate, evolve), helping people do what they do best (create, lead, judge, improvise).
In a Harvard Business Review webinar on this subject, Wilson offers the example of a healthcare provider whose team of radiologists and pathologists created an AI-based technique to identify breast cancer cells. In testing that technology, the team found that the doctors outperformed the machine, delivering 96% accuracy compared with 92%. But when the doctors and AI combined forces, they accurately identified 99.5% of cancerous biopsies. That increased precision, Wilson says, would translate to about 130,000 more women every year receiving accurate breast cancer diagnoses than if the healthcare system relied on doctors or intelligent machines alone.
Introduced correctly, autonomous technology has potential to lower costs, improve quality, and increase access to healthcare. But for all this to happen, Abramoff says, “we need to do it right.”
— Margaret Lindquist
At the center of each of those signals are databases that hold, manage, and protect that data. And behind those databases are the administrators who set them up, back them up, protect them against hackers, and fine-tune their performance.
Like people in many industries, the job of the database administrator, or DBA, will change as autonomous technology moves into their workdays. The first-of-its-kind “self-driving” autonomous database from Oracle, which builds on the company’s decades of engineering work to automate tasks for DBAs, now adds machine learning to deploy, tune, patch, upgrade, secure, and back itself up with no human intervention.
Such changes can seem scary to longtime DBAs, says Afredo Abate, a DBA and systems architect for Brake Parts Inc. and also the president of an Oracle technology user group in Chicago. “And I’ll be honest: I was scared at first, too,” he says. “They’re talking totally autonomous, and it’s going to do all these DBA-ish things for you. What does this mean for me?”
Abate’s user group met to test Oracle Autonomous Database for themselves. “Click, click, click, and there’s your database environment,” he says. “You’re done. Whereas before you had to download software and get it all up and running, that portion is done.” Now developers or data analysts “can just log in and spin up their own database” without asking a DBA for help, he says.
Abate’s fear has been replaced by intrigue at the new opportunities for people who understand data. “There are other things we DBAs offer,” he says.
For example, Abate has taken it upon himself to learn how data is used throughout the company’s brake factory and financial operations. “As a DBA you can help people understand how to get data to and from all these applications and systems,” he says. “Now a key skill set for people like us is going to be moving data to the cloud and back again.”
The autonomous database “is only going to get smarter,” Abate says, but he sees that as an opportunity to do more strategic work. “I have a list of things that I want to get done for our business—I have ideas,” he says. “But I can’t get to them because I’ve been too busy doing mundane tasks.” The autonomous database has arrived just in time.
—Jeff Erickson
Agencies in the US and Switzerland think it can and, with the help of researchers at several universities, they’re testing software to make those life-changing decisions.
One such researcher is Andrew Trapp, a professor at Worcester Polytechnic Institute in Massachusetts, who’s part of an international team of data scientists and economists who’ve developed a machine learning algorithm that predicts refugee employment probabilities, using optimization models to recommend the best possible matches to help agencies resettle thousands of displaced people into US communities.
They’ve name the program Annie MOORE (Matching and Outcome Optimization for Refugee Empowerment), after the first immigrant to enter the US through New York’s Ellis Island in 1892. The algorithm is fed US State Department data, such as the refugee’s nationality, language proficiencies, and ages of family members, along with community data such as open slots for new refugees, whether there is support for single parents or large families.
In 2018, HIAS began using Annie to help resettle refugees. While HIAS staff still have the final say about where refugees are placed, the nonprofit says its staff now spends less time on routine matching and focuses instead on more difficult case work, such as placing people with serious medical conditions.
Government isn’t often cited as a common use case for autonomous technology: Just 28% of survey respondents think we’ll routinely see autonomous tech used by government agencies in the next five years. Yet when we asked how soon we’ll see the technology routinely making office work decisions, such as approving expense reports and giving employees IT access, 67% say that will happen in four years or less.
The Swiss government, working with researchers at the Immigration Policy Lab (IPL), which has branches at Stanford University and ETH Zurich, is also using machine learning to help place refugees. The Swiss government is piloting IPL’s placement algorithm, making it the first country to place asylum seekers in different parts of the country using artificial intelligence, IPL says.
The IPL technology continually learns based on experience and changing conditions, using data from the Swiss State Secretariat for Migration. That data includes details about where refugees were sent, whether they found work, their age, country of origin, gender, and time of arrival.
“We’re training our algorithm to look for patterns in data about when and where former asylum seekers found jobs,” says Joëlle Pianzola, IPL’s executive director at ETH Zurich, a science and technology university founded in 1855. “When a new asylum seeker with similar characteristics comes to Switzerland, the algorithm can instantly make a recommendation as to where the best place is for her to be.”
As with Annie’s use in the US, Swiss caseworkers can override the algorithm’s choice. “This is still human-centered artificial intelligence,” Pianzola says. “This is about complementing human decision-making, not replacing it.”
—Sasha Banks-Louie
