Artificial intelligence (“AI”) is a fast-growing technology that replicates human cognitive abilities by analyzing billions of pieces of information in only seconds, and producing new content in the form of texts, images, audio, and more. AI has expanded worldwide and become a popular tool in one’s day-to-day life. AI has also revolutionized the workplace by streamlining productivity and efficiency. Below is a discussion of how AI has been implemented in the workplace and the benefits it has produced, while also highlighting the risks and pitfalls that come along with using AI.
AI in Everyday Life
AI has actually been around for a long time and individuals use AI every single day without even realizing it. Unlocking your phone using face ID, sending an email corrected by editing tools, googling a question and receiving thousands of results, communicating with Siri or Amazon Alexa, using a GPS, or browsing the internet and seeing advertisements and shopping recommendations that are specialized to your personal style, all require AI. The purpose of AI is to make our lives easier by eliminating tedious tasks, virtually assisting us, personalizing content we see online, and much more.
AI in the Workplace
Utilizing AI to complete mundane tasks in everyday life is not the only way AI has been implemented. A study by the National Bureau of Economic Research found that the use of AI in the workplace led to an increase in customer issues solved per hour, a reduction of managerial involvement, and positive customer sentiments. By utilizing AI in the workplace, employers can optimize their time and productivity much quicker than ever before.
Automating Repetitive Tasks
AI allows for automation, meaning repetitive tasks such as checking your calendar, scheduling meetings and appointments, processing payroll, responding to client/customer service requests, and more, can be done with automated AI. Employers can use virtual assistants and chatbots on their websites to handle client/customer inquiries faster. By combining Robotic Processing Automation software with AI, repetitive tasks such as data entry are accomplished faster with a lower risk of human error. Automated responses to emails and questions could be produced by combining AI with Natural Language Processing systems. With automated AI services, employers are dramatically cutting down on the time and cost to complete administrative tasks.
Enhancing Decision-Making
With the ability to analyze billions of pieces of data at a time, AI allows employers to gain better insight on trends and patterns in their field of work. This can be done by having AI analyze historical data and make predictions, or it can offer real-time insight that can help an employer make short-term and long-term decisions. AI may help employers manage risks by identifying and assessing them. To strengthen and quicken the decision-making process, employers have turned to AI to analyze information that the human mind may not be able to accomplish as quickly or as accurately.
Recruiting Employees
AI has also taken the stress out of hiring employees by revolutionizing the recruitment process. AI can review resumes and cover letters submitted by candidates and detect keywords that AI believes best aligns with the job. This can make finding qualified candidates easier. AI can also update job descriptions, so the appropriate candidates are applying for the job. By quickening the application process, employers can be more efficient in hiring employees, thus securing a sturdier workforce.
The Pitfalls of AI in the Workplace
With every beneficial aspect associated with AI, there are downsides to it as well. AI is hardly perfect and is constantly developing, thus requiring careful human oversight. AI is not meant to replace human cognitive abilities. Instead, AI may best be used as a collaborative tool that augments human capabilities. There are various risks that employers should be wary of when deciding to implement AI in their workplace.
Hallucinations
AI may be prone to producing something known as “hallucinations.” Hallucinations are incorrect answers that AI generates in response to a prompt, which can mislead a user. AI may produce a comprehensive report with seemingly legitimate citations and sources, but those citations and sources never existed in the first place. This problem is especially prevalent in the academic and scientific field. AI is trained to work with the data it is provided; if that data is false, misleading, or insufficient, AI may fill in the missing pieces of information with erroneous information. When using AI, oversight and doublechecking accuracy is critical. The more comprehensive, relevant, and accurate the data is, the better the results.
Accountability
Accountability issues are also raised when using AI. To ensure that employers are compliant with state law and regulations such as anti-discriminatory laws, or that the information AI is generating is correct, human accountability is required. This may mean employing risk management techniques as well as guidelines on how AI is to be used in the workplace, what uses are prohibited, and how to better train AI to analyze information. When working with AI leads to unplanned events, it is important to establish accountability to maintain trust with clients/customers and employees.
Transparency
When there is a lack of plausible explanations to rationalize and decipher how certain decisions were made, transparency becomes an issue. If AI comes to a conclusion on a particular issue, an employer relying on that AI generated answer may want to ensure they can articulate why they trusted that AI generated recommendation. This may require explaining why AI was implemented in the first place, what information AI processed, what tools AI used in its analysis, and how human oversight was conducted. These disclosures can also make holding individuals accountable easier by requiring a comprehensive interpretation and understanding of what AI is doing when it is utilized.
Bias
Bias is a serious risk associated with AI that employers may want to take into consideration. Unintentionally, AI may discriminate against protected classes of people. AI is notorious for supposedly being unbiased and this may be true when AI is used correctly. However, since AI makes decisions based on the data it is provided, if the human produced data that AI is fed contains bias, AI may train itself to pick up on this bias.
In the hiring area, AI must be carefully planned, to avoid mistakes or discrimination against protected classes of people. For example, Amazon previously used a hiring algorithm to sift through resumes to hire software developers. After discovering that their AI recruitment system favored male applicants over female applicants, Amazon disbanded the AI system. AI had trained itself to associate certain words commonly found in a male’s resume more favorably than a female’s resume.
In 2023, New York City’s Local Law 144 took effect, which prohibited the use of Automated Employment Decision Tools (AEDT), unless the employer conducts a bias audit and properly notifies a job candidate that AEDT will be used. AEDT includes AI and other technologies such as learning machines or data analytics. For employers who use AEDT to help them make employment decisions and the AEDT substantially assisted the employer, a bias audit must be conducted by an impartial independent auditor. The audit must include the selection rate for categories such as sex and race/ethnicity, and the impact ratio of each category. California has also proposed their own comprehensive bill to regulate AI to prevent discrimination in areas such as employment, education, housing and more.
On a federal level, the Biden Administration has published a guidance called the Blueprint for an AI Bill of Rights (Blueprint). The Blueprint is meant to act as a guide on how to design and use AI to protect society from the unequal treatment that AI can create, with employment being one of the concerns raised in the Blueprint. The Equal Employment Opportunity Commission (EEOC) has also taken the initiative and issued their own guidance to ensure that AI used in the employment decision making process does not violate federal civil rights laws, with resources on how to prevent bias and discrimination periodically released by the EEOC.
Takeaway
AI has been around for a long time, and it is here to stay. AI has managed to bring efficiency not just to the lives of individuals using AI every day, but also to the employers and employees in managing their day-to-day operations. While implementing AI in the workplace can bring great benefits to an employer, being cautious of the risks associated with AI is critical if AI is to be used correctly and safely in the workplace. We assist our clients in drafting their written policy on AI. Overseeing the output AI generates could potentially avoid issues such as bias, hallucinations, accountability, and more. Instilling proper guidelines and oversight will ensure that AI is used appropriately.