Artur Kiulian

@arturkiulian

Why Your Next Boss Will Be A Robot

August 8th 2017

Thoughts On How Likely Your Work Will Be Judged By A Machine

Artificial intelligence software and robots are powerful in pattern recognition, predictive analytics, heavy computations, and handling repetitive tasks. Thanks to these capabilities, machines are gradually replacing humans in many occupations and activities, to extent of a growing concern about the impact of automation on the job market.

“A lot of valuable work currently done by humans — examining security video to detect suspicious behaviors, deciding if a car is about to hit a pedestrian, finding and eliminating abusive online posts — can be done in less than one second. These tasks are ripe for automation. However, they often fit into a larger context or business process; figuring out these linkages to the rest of your business is also important” — says Andrew Ng, who recently announced his $150m AI focused venture capital firm

While the power of AI is indisputable, the question arises how far will automation go and what will be its impact on employees, organizations, and business processes. The main question is — will AI become the next boss for the majority of employees?

Most experts agree that the majority of occupations will be partly or fully automated in the near future. In practice, this means that employees will be either fully replaced by machines, or begin working with them as assistants, trainers, or subordinates. Some predictions made by AI experts are fairly radical.

For example, McKinsey & Company came to the conclusion that currently available AI technology could automate around 45 percent of all activities people perform today. The consultancy company also predicts that around 60 occupations will experience at least 30 percent of automation in the coming decades.

Automation Scenarios

The scope of automation will depend on the technical feasibility (percentage of time spent by employees on activities that can be automated), regulatory preparedness and social acceptance, economic benefits and costs of automation.

McKinsey & Company argues that managing others and applying expertise are activities the least susceptible to automation. More susceptible are tasks that involve stakeholder interactions and unpredictable physical work (e.g construction, forestry). Finally, the most susceptible to automation are predictable manual work, data collection, and data processing.

Intuitively, robots can become your next boss if you stay employed. Therefore, in such spheres as autonomous driving or automated trading machines will be their own bosses. In autonomous driving, for example, a truck fleet movement will be synchronized via over-the-air systems (OTA) that can instantly send software updates and security fixes in case of emergency. Obviously, no serious human intervention is needed for such systems to work smoothly.

New forms of human-machine communication, including AI-guided management, are more feasible in the hybrid companies where AI and employees are integrated into the coherent business process.

The first scenario here are machines that assume the role of customer-facing interfaces that facilitate company operations, such as customer support or business analytics. Vertical chatbots that perform one specific task of connecting consumers to human consultants is a typical example of this model.

Many startups hire so-called chatbot trainers that evaluate the performance of chatbots and step in when something goes wrong. In this model, human employees augment and assist AI software leveraging its natural language processing (NLG), analytics, image recognition or other ML functionality to run business processes and make important decisions. In the end, however, it is human managers and employees who have the upper hand.

So, When Robots Will Become Your Boss?

AI is most likely to become robobosses in companies that heavily rely on algorithmic solutions in their decision-making or whenever machines participate in managing employees and evaluating their performance. According to Gartner, by 2018 3 millions of people globally will be supervised by robots.

Robobosses are already performing important tasks in many data-driven companies. For example, the world’s largest hedge fund Bridgewater Associates that oversees over $160 billion is building the PriOS algorithmic management system that controls all basic business processes and operations. The system is responsible for a number of managing tasks such as hiring and firing employees or ranking opposing perspectives to solve disputes and disagreements in the team. The rationale behind this AI system is to fully exclude any impact of emotions and mood on investment decisions.

AI software is also gaining traction in banking and mortgage brokering where machines decide what clients are eligible for lending. Without such software, mortgage brokers would spend 90 percent of their time reviewing applications.

The machine can do it more efficiently freeing up time for brokers to consult and advise clients. There is growing concern, however, that hidden biases discriminating minorities might creep into ML algorithms used in the credit scoring software.

Companies can also leverage the power of complex image recognition software to automatically assess the performance of employees in contexts where it can not be properly measured by human supervisors.

For example, robobosses can track the wheel angle chosen by the Uber driver to evaluate his or her driving skills and style. Similarly, in the service sector ML algorithms can be used to evaluate how well a hostess or a waiter do their job by tracking the percentage of completeness their smile reaches in front of the VIP clients.

Algorithms Are Already Managing People

and we are not paying much attention to that fact

AI software is in charge of important business decisions, planning and performance assessment in many on-demand mobility and delivery services that make up the so called gig economy. In Deliveroo, a London-based food delivery company, most of the couriers’ actions are tightly controlled by algorithmic management.

If a courier declines the order, strict algorithms would penalize them. Deliveroo’s algorithmic system carefully monitors a courier’s performance calculating his/her average “time to accept orders”, “travel time”, and “unassigned orders”. If the courier’s performance does not meet a service level agreement, he or she might be blocked in the system.

Similar algorithmic procedures are used in Uber, the world-leading mobility service that connects passengers and drivers. At Uber, once a driver is logged into the system he or she has 10–20 second to accept trip requests. If three trip requests are missed in a row, a driver is automatically logged out for several minutes. In the case of frequent violation of Uber algorithmic policies, the driver’s account may be deactivated.

Along with these algorithms, Uber drivers raised complaints against Uber’s Dynamic Pricing Model. It automatically sets fares depending on the aggregate demand on Uber services. As a result, Uber drivers’ earnings are highly unstable. Uber’s unjust price setting policies caused drivers to join “Fight for $15” protests in 2016 in which drivers demanded a fair pay for their services, union rights, and social benefits.

As these examples illustrate, machines are likely to become bosses in businesses that depend on data-driven systems, automated analytics, and algorithmic decision-making. Whenever the company transfers responsibility for decision-making to ML algorithms, machines automatically turn into bosses.

At the same time, however, robot bosses will currently not affect the spheres where wrong decisions have dangerous implications.

This is true about health care in which daily activities of health professional require human expertise and direct contacts with patients. Although medical diagnosis AI can provide physicians with valuable insights, the final decision will be taken by health care specialists who have to verify the machine’s conclusions and ensure that all ethical norms and policies are met.

Wondering how likely your profession is a subject to algorithmic management? Check out the tool I’ve built using the public work activity and occupation data along with the Oxford computerization research:

More by Artur Kiulian

More Related Stories