Employers and Artificial Intelligence: What Should We Know?

The growth of artificial intelligence-based tools gives employers the opportunity to increase efficiency, and to save time, money, and energy in historically human-performed tasks.  One of today’s fastest developing AI tools is called Chat GPT, an artificial intelligence chatbot that generates human-like responses to the questions and instructions it is given. Chat GPT collects data from throughout the internet and uses computing predictions to answer prompts input by the user. The tool, like most artificial intelligence, learns about subjects through its own data collection and replies, and then develops its ability to communicate about those subjects with practice. While many AI technologies are still considered complex emerging technologies, Chat GPT is accessible to any individual simply by establishing a free account on its website. Such convenient access allows people in all areas of employment to utilize Chat GPT in their everyday work and to implement the power of AI into their companies.

For example, Chat GPT can streamline traditionally human-performed employer tasks such as screening applicants and deciding when promotions, layoffs, and terminations are appropriate. Internally, Chat GPT can help employees more efficiently prepare documents, write summaries, enhance research, and countless other assignments. As the use of artificial intelligence booms, many countries and cities are implementing stricter guidelines on its use. New York City recently issued Local Law 144, which requires an audit on bias when an employer uses AI software in making employment decisions. Similarly, companies such as Amazon, JPMorgan, Bank of America, and Goldman Sachs have banned the internal use of Chat GPT, citing privacy concerns. While Chat GPT can provide incredibly impressive and helpful assistance, employers should be aware of the possible risks it can present.

1. Bias. Chat GPT learns from the data that it is trained on, and if that data contains biases, Chat GPT may perpetuate them in its answers.  Any bias in the training data, including race, gender, age, and disability biases, can result in biased responses against candidates and employees. As every employer knows, Title VII of the Civil Rights Act of 1964 prohibits discrimination on the basis of race, color, religion, or national origin; the Americans with Disabilities Act (ADA) prohibits discrimination against qualified individuals with disabilities; and the Age Discrimination in Employment Act prohibits age discrimination. Artificial intelligence, however, has produced responses that do not take these laws into account. For example, in 2018, Amazon discontinued its highly anticipated AI recruitment technology when it appeared to discriminate against female applicants. In recent years the Equal Employment Opportunity Commission (EEOC) has focused heavily on preventing AI from discriminating in employment-based decisions. In 2021, the EEOC launched its Artificial Intelligence and Algorithmic Fairness Initiative, which attempts to ensure that any AI used in employment decisions complies with all federal civil rights laws.  And in 2022, the EEOC published guidance on the use of AI in employment decisions and compliance with the ADA.  (https://www.eeoc.gov/laws/guidance/americans-disabilities-act-and-use-software-algorithms-and-artificial-intelligence.) Even with this progress, employers must continue to ensure that the training data that Chat GPT uses is representative of the diverse pool of candidates at issue and that it has not learned bias against any legally protected group. Failure to do so can result in discriminatory decisions and open employers to legal exposure.
2. Privacy. All information entered into the Chat GPT prompt box has the potential to be retrieved by third parties, AI trainers, or even incorporated into the Chat GPT’s model, and should not be considered secure. Any resulting data leaks can have extremely serious consequences for both organizations and individuals. For companies, data leaks could result in loss of intellectual property, release of privileged information, and damage to reputation, among other results. For individuals, data leaks could lead to financial loss and identity theft. While Chat GPT has implemented measures to protect against such security breaches, the technology is new enough that the accuracy of these measures is still being tested. It is vital that employers put guidelines in place to ensure that employees are not inadvertently exposing themselves and their organizations to these privacy risks.
3. Accuracy. Chat GPT’s accuracy may vary depending on the quality of its training data and the complexity of the task it is given. According to Chat GPT’s website, the current model has limited knowledge of the world and events after the year 2021 and any responses should be fact-checked by humans. While Chat GPT tries to provide accurate information, it often gets things wrong because it depends on patterns in data rather than on facts. Employees and employers should be cautious in depending on Chat GPT for accurate, unflawed work product, as the technology’s accuracy is still developing. To err on the side of caution, employers should always double check the accuracy of Chat GPT’s work before relying on it.
4. Legal Compliance. As the use of AI spreads, many companies, and even some cities and countries, have banned the use of artificial intelligence tools in areas of employment and use on the job. Before utilizing these tools in the workplace, employers should consider all the compliance requirements they may be subject to. To avoid any legal violations, all organizations should confirm that the use of AI is allowed and that any rules surrounding such use are being followed.