The Direction of Smart Automation

Published on INFORMS Analytics Magazine (Joseph Byrum)

What humanity optimized by artificial intelligence might look like

Tomorrow’s most vexing problems will be solved with smart automation. By combining key technologies – artificial intelligence (AI), big data, advanced analytics and autonomous systems – smart technology can achieve far more than the unassisted human mind could ever accomplish.

Smart automation represents the next frontier in improving the human condition, but it has been in the works for decades. The automotive industry, for instance, was an early adopter of increasingly smart automation.

In the 1970s, General Motors came up with CONSIGHT, a primitive form of machine vision that used a camera to allow a robotic arm to “see” objects on a conveyor belt, pick them up and deliver them to another work area [1]. Amazing for its time, but now consumers can drop by a dealership and pick up an AI-powered, semi-autonomous Tesla car that can see obstacles and react better than a human can. The manufacturer claims, with government stats to back it up, the Tesla will be 40 percent less likely to crash than a human-piloted car [2]. That’s a rather remarkable advance in just a few decades.

No wonder three out of four managers surveyed by the consulting firm Infosys said AI is “fundamental” to their organization’s success [3]. They see the evidence continuing to mount that smart technology can outperform humans in a number of tasks. One MIT study [4] found machines were better not just at analyzing data, but they could better decide what data to analyze in the first place. The machines proved they could do in a matter of hours what the best human teams took months to accomplish.

Another study [5] from researchers at Yale and Oxford Universities looked at the question of how long it would take for smart technology to surpass human abilities across a broad variety of fields, including a few areas that no one would have imagined could ever be taken over by a machine operating without direct human guidance. The consensus of 352 expert practitioners in the AI field was that machines will beat human translators by 2024, they’ll surpass the brightest high schoolers at essay writing by 2027, and they’ll be churning out bestseller-quality novels by 2049.

A few years later, the study reports, smart automation will replace human surgeons (already, machines perform highly complex laser eye surgery). AI researchers believe that their own field, AI research, will be the last to go, replaced by machine intelligence some 80 years from now.

By combining key technologies – AI, big data, advanced analytics, autonomous systems – smart technology can achieve far more than the unassisted human mind could ever accomplish. Photo Courtesy of ThinkStock.com

As we head into this future, it’s worth reflecting on how humanity will change as smart automation takes on a greater role in our lives. As with all human endeavors, taking the time to pause and reflect on the balance of what is gained and what is lost in this newest revolution can pay off.

The most obvious drawback to smart automation is job loss and the attendant personal, social and economic strains from unemployment. At each stage of development – from early machines to the present –humans have ceded more and more control to external, autonomous systems. In doing so, we humans have also relinquished control over our own data and, to some degree, destiny. In this new era, we are witnessing many changes including even the emergence of what could be called the human machine.

The Human Machine

Earlier this year, the U.S. Congress began a series of hearings investigating how the private and public sectors should prepare for the impact of AI on society [6]. The European Union’s own deliberations last year explored the interesting question of who should take the blame when a robot injures someone. The parliamentary committee suggested [7] that a human ought to be held responsible to the extent that person influenced the robot’s actions or skills. “Skills resulting from ‘training’ given to a robot,” the committee report explained, “should not be confused with skills depending strictly on its self-learning abilities when seeking to identify the person to whom the robot’s harmful behavior is actually attributable …” For now, the committee was comfortable finding some human to take the blame.

Yet the European inquiry did not end there. The committee considered whether it makes sense to create a “specific legal status for robots.” The idea is to give “the most sophisticated autonomous robots … the status of electronic persons responsible for making good any damage they may cause, and possibly applying electronic personality to cases where robots make autonomous decisions or otherwise interact with third parties independently.”

That seems like something only necessary in a science fiction thriller, but the future may not be so far away. As we’ve seen, the capabilities of smart automation are growing fast, and as each developmental milestone is surpassed, the notion of autonomous intelligence becomes less far-fetched. For this reason, it is essential that work in smart automation be firmly rooted in ethical guidelines, such as those currently being developed by a number of professional and legal communities.

More than a quarter century ago, the Association for Computing Machinery began with an ethics code for the industry [8]. The Institute of Electrical and Electronic Engineers (IEEE) has also issued relevant standards, with the most recent work, “Ethically Aligned Design: A Vision for Prioritizing Human Well-being with Artificial Intelligence and Autonomous Systems” [9], offering guidance as smart automation projects become more sophisticated. The UK’s Engineering and Physical Sciences Research Council came up with “Principles of Robotics” [10] to bring Asimov’s laws into the 21st century.

According to Infosys, 62 percent of the senior decision-makers surveyed agreed that stringent ethical standards are needed to ensure the success of AI deployments, but only a third reported that their organizations had “completely considered the ethical issues relating to the use of AI” [11]. There are a number of organizations looking to change that.

INFORMS has its own College on Artificial Intelligence [12] that has held workshops on the uses of AI to solve business challenges and the societal impact this can bring about. Corporate giants such as IBM, Amazon, Google, Facebook and Microsoft, along with partner organizations and non-profits, have created the Partnership on AI to Benefit People and Society [13] to broaden the dialogue about smart technologies. The very title of the organization embodies a positive message focused on education and public engagement.

Many organizations are creating industry-specific, practical rules for their membership to follow. The American Farm Bureau, for instance, has come up with a practical set of privacy and security principles for farm data [14] that leading industry groups have agreed to follow. “We believe farmers own information generated on their farming operation,” the document states. The signatories agree to only use and collect this valuable data with the express consent of the farmer, showing how arrangements can be made to the mutual benefit of all parties involved.

Optimizing Humanity Itself

Building a solid ethical foundation for smart automation becomes even more critical as smart automation begins toying with the possibility of altering humanity itself. Yuval Noah Harari, author of the bestselling books, “Sapiens: A Brief History of Humankind” and “Homo Deus: A Brief History of Tomorrow,” offers many important insights. He points out that humans have radically transformed their environment through the agricultural, industrial and information revolutions, yet over the course of 10,000 years, mankind has not changed. Harari predicts the next revolution in human existence will be of a different kind, as it will change humans themselves. Biotech and artificial intelligence will combine to alter our physical and cognitive abilities, as well as our very biology.

The development of brain-computer interfaces, or brain-machine interfaces, has restored mobility to the paralyzed through implants that give the brain the ability to operate a prosthetic limb [15]. Duke University neuroscientist Mikhail Lebedev, who works on such devices, believes that one day brain augmentation will enable our minds to interface directly with AI. This opens unlimited potential for expanding our sensory abilities. Imagine being able to perceive the electromagnetic spectrum, or to have 360-degree vision.

This may also seem far-fetched considering how complex our brain is, but it is perhaps not too complex for science to reconstruct for the purpose of expansion and alteration. Our neural circuitry enables our thoughts and behavior, and a full understanding of how it works is within reach of future generations, according to Harvard’s Connectome project. “One hundred years from now, this brain circuitry will be known; today, for the first time, we can contemplate mapping it in detail,” the project claims. “New forms of laser-scanning light microscopy and semi-automated electron microscopy allow high resolution imaging of ‘connectomes’—that is, complete neural wiring diagrams” [16].

So much of what smart automation can do today would have seemed like magic not so long ago. That’s why it’s important to think not in terms of what it can’t do, but to prepare for what might be possible.

The potential for the smart automation of the future to optimize humans is on a scale far beyond merely improving business processes or constructing higher-quality automobiles. While we have the luxury of time to debate and explore the implications, it falls to us, in the early stages of smart automation, to agree upon a solid set of ethical guidelines. Doing so will help direct research efforts and innovation toward the projects that truly enhance human well-being.

References

1. https://link.springer.com/chapter/10.1007/978-1-4613-3027-1_5
2. https://www.theverge.com/2017/1/19/14326258/teslas-crash-rate-dropped-40-percent-after-autopilot-was-installed-feds-say
3. https://www.infosys.com/aimaturity/Documents/amplifying-human-potential-CEO-report.pdf
4. http://www.eecs.mit.edu/news-events/media/automating-big-data-analysis
5. https://arxiv.org/pdf/1705.08807.pdf
6. https://oversight.house.gov/hearing/game-changers-artificial-intelligence-part-ii-artificial-intelligence-federal-government/
7. http://www.europarl.europa.eu/sides/getDoc.do?pubRef=-//EP//TEXT+REPORT+A8-2017-0005+0+DOC+XML+V0//EN
8. https://www.acm.org/about-acm/acm-code-of-ethics-and-professional-conduct
9. http://standards.ieee.org/develop/indconn/ec/ead_v1.pdf
10. https://www.epsrc.ac.uk/research/ourportfolio/themes/engineering/activities/principlesofrobotics/
11. https://www.infosys.com/aimaturity/Documents/amplifying-human-potential-CEO-report.pdf
12. http://connect.informs.org/cai/home
13. https://www.partnershiponai.org/
14. http://www.fb.org/issues/technology/data-privacy/privacy-and-security-principles-for-farm-data
15. http://scipol.duke.edu/content/brain-computer-interface-based-neuro-prosthetics
16. http://cbs.fas.harvard.edu/science/connectome-project