The advent of the pandemic has resulted in the virtualization of many jobs, perhaps permanently. The associated increased quantification of the workplace will enable the accelerated adoption of AI. As our private and professional lives inevitably blur as we work more from home, we need ethical rules more than ever about how companies can use employee-generated data and deploy AI in the workplace.
Increased virtualization is likely permanent for many jobs, paving the way for workplace bots
One lasting change from COVID-19 is likely to be that we all will be working from home (or anywhere) more often. Many companies have announced they will allow work from home as a permanent policy. According to a recent survey, most managers have been pleasantly surprised by the change, and many have seen a marked increase in productivity from their employees. The success of this rapid adaptation to virtual work is largely due to the technology tools available, including some AI-enabled ones, which allow managers and administrators to monitor and maintain control remotely.
This shift has big implications for the acceleration of AI adoption in the workplace. The virtualization of the workplace means more data, and more data means more opportunities for effective machine learning. It’s worth noting that data collected in the process of work may generally be used by employers without the consent of employees. There is also an increased risk that our personal data may get mixed in and become visible to employers and their bots as our home and office life converge.
AI is already used to manage employee performance
AI is already being used in the workplace to evaluate and improve employee performance. For example, AI is being used to provide customized training to employees realtime. Companies like Chorus monitor sales calls and communications to give tailored feedback and measure performance improvement. Another is Cogito, which is a tool that combines AI with behavioral science to help customer service employees provide better phone support through augmented emotional intelligence. Managers can be automatically flagged to intervene on calls that are going badly.
If this all seems creepy, things are likely to get worse. With an increasingly virtual workplace, even more of an employees’ performance will be visible to an AI to evaluate. Feedback from an AI may help people build skills and become more proficient at their jobs. At the same time, it will allow companies to squeeze more productivity from their workers. This will undoubtedly lead to higher performing companies, but working in this digital fishbowl may be unnerving for some.
Say hello to your AI admin (or boss)
AI can help with administration functions in the workplace, from taking notes to managing projects. With speech-to-text nearly as good as humans, AI is able to act as notetaker during virtual meetings. One startup doing this is Otter.ai, which allows Zoom meetings to be transcribed.
AI may perform some admin tasks that are superhuman. Google Meet, another video chat service, generates real-time closed captions as an option. Skype uses AI to automatically translate for both parties in a conference call. Plus, with everything said being transcribed, the text can be fed into other AI systems for clustering topics, summarizing discussion, capturing decisions and pulling out next steps and assignments.
Having an AI admin in the virtual workplace has many obvious business advantages beyond note taking, potentially increasing meeting effectiveness and knowledge capture. At the same time, having an AI listen in may lead to a lack of candor in the virtual workplace, or worse yet, it may breach the privacy and confidentiality of employees. These risks are especially concerning due to the inevitable mixing of our private and professional lives.
An ever-present AI watchdog may foster a more fair work environment
On a more positive note, AI has the potential to make the virtual workplace more welcoming and fair. It’s quite possible that the virtualization of work and the application of AI may reduce discrimination and sexual harassment. Managers may be on better behavior simply because their words and actions could be observed or recorded by someone in a virtual meeting. AI has already been used to monitor employee behavior to comply with corporate policies against bullying and harassment by reading employee emails and instant messages. Now the content of virtual meetings could be added to the mix. In the virtual work environment, bots may be welcome members of the team.
Virtualization and AI together give more power to employers
Nonetheless, all of this virtualization and application of AI is upending the balance of power towards employers, and away from employees (and ostensible contractors). Gig workers were already especially vulnerable to AI. Self-driving cars are the most visible looming threat to Uber and Lyft drivers. But such total replacement of jobs is still a ways off. Instead, AI is chipping away at the task level of many jobs, especially knowledge workers. The recent increased virtualization of those roles will only accelerate the process of AI augmentation. If employees are already remote and reduced to a stream of data (in the form of emails and videos), then the effort to replace them with an AI becomes more trivial.
Use of AI in the increasingly virtual workplace needs to be regulated more than ever
Some employee personal information, including SSN and health information, is protected by law. But data collected and generated by an employee as part of their work generally remains the property of the employer, who may use it freely, including feeding it to AI systems.
Here are five principles for the ethical use of AI in the workplace
Safeguard employee data
Be transparent about the use of AI with employees
Guard against bias and unintended consequences
Provide explanations of machine-made decisions that affect employees
Adopt a broad definition of employee
I raised many of these concerns in an interview over two years ago. The potential for abuse is even greater now due to the increased virtualization and blurring of our home and work lives. Laws and business practices need to catch up to the new reality.
Note: This article is a followup to an interview I gave to Think magazine two years ago about the ethics of AI in the workplace. It is based on a lecture given by me at UC Berkeley, Haas School of Business, as part of the “Business of AI” online course live session on June 2, 2020.