AI can be a helpful employee management tool — and a legal liability in decision making

Skip to Content News Archives Economy Energy Oil & Gas Renewables Electric Vehicles Mining Commodities Agriculture Real Estate Mortgages Mortgage Rates Finance Banking Insurance Fintech Cryptocurrency Work Wealth Smart Money Wealth Management Investor Personal Finance Family Finance Retirement Taxes High Net Worth FP Comment Executive Women Puzzmo Newsletters Financial Times Business Essentials More Innovation Information Technology FP500 Podcasts Small Business Lives Told Tails Told Shopping Financial Post Store Obituaries Place a Notice Advertising Advertising With Us Advertising Solutions Postmedia Ad Manager Sponsorship Requests Classifieds Place a Classifieds ad Working Profile Settings My Subscriptions My Offers Newsletters Customer Service FAQ News Economy Energy Mining Real Estate Finance Work Wealth Investor FP Comment Executive Women Puzzmo Newsletters Financial Times Business Essentials This advertisement has not loaded yet, but your article continues below.HomeLegal PostWorkAI can be a helpful employee management tool — and a legal liability in decision makingHoward Levitt: Using AI does not transfer legal responsibility from the employer to the machineEmployers are rapidly adopting systems that rank applicants, assess performance, identify unusual employee behaviour and recommend management action. Photo by Getty Images/iStock PhotoThe most dangerous artificial intelligence decision in your workplace may be the one no one can explain.THIS CONTENT IS RESERVED FOR SUBSCRIBERS ONLYSubscribe now to read the latest news in your city and across Canada.Exclusive articles from Barbara Shecter, Joe O'Connor, Gabriel Friedman, and others.Daily content from Financial Times, the world's leading global business publication.Unlimited online access to read articles from Financial Post, National Post and 15 news sites across Canada with one account.National Post ePaper, an electronic replica of the print edition to view on any device, share and comment on.Daily puzzles, including the New York Times Crossword.SUBSCRIBE TO UNLOCK MORE ARTICLESSubscribe now to read the latest news in your city and across Canada.Exclusive articles from Barbara Shecter, Joe O'Connor, Gabriel Friedman and others.Daily content from Financial Times, the world's leading global business publication.Unlimited online access to read articles from Financial Post, National Post and 15 news sites across Canada with one account.National Post ePaper, an electronic replica of the print edition to view on any device, share and comment on.Daily puzzles, including the New York Times Crossword.REGISTER / SIGN IN TO UNLOCK MORE ARTICLESCreate an account or sign in to continue with your reading experience.Access articles from across Canada with one account.Share your thoughts and join the conversation in the comments.Enjoy additional articles per month.Get email updates from your favourite authors.THIS ARTICLE IS FREE TO READ REGISTER TO UNLOCK.Create an account or sign in to continue with your reading experience.Access articles from across Canada with one accountShare your thoughts and join the conversation in the commentsEnjoy additional articles per monthGet email updates from your favourite authorsSign In or Create an AccountEmployers are rapidly adopting systems that rank applicants, assess performance, identify unusual employee behaviour and recommend management action. The attraction is obvious. AI can process vastly more information than a manager, almost instantaneously.This advertisement has not loaded yet, but your article continues below.But using AI does not transfer legal responsibility from the employer to the machine. If it contributes to a hiring, discipline, accommodation, promotion or termination decision, the employer remains responsible for complying with the law. The critical question is how much the AI influenced the decision and to what extent the employer critiqued the analysis before implementing its conclusions.FP Work touches on HR strategy, labour economics, office culture, technology and more.By signing up you consent to receive the above newsletter from Postmedia Network Inc.A welcome email is on its way. If you don't see it, please check your junk folder.The next issue of Work will soon be in your inbox.We encountered an issue signing you up. Please try againAI materially influences an employment decision when its output plays a meaningful role in determining what the employer does. It might narrow the number of candidates considered for a job, flag an employee for investigation, generate a score management relies on or recommend a course of action.That is different from using AI to draft a job advertisement, summarize a meeting or perform some other administrative task that does not meaningfully affect the decision.The trouble begins when a manager stops using AI to help with the work and starts using it to decide what the work means.Consider an employer that uses AI to identify employees whose performance supposedly warrants further scrutiny. An employee is repeatedly flagged. The manager investigates and ultimately terminates them.This advertisement has not loaded yet, but your article continues below.Months later, in litigation, the employee’s lawyer asks:“Why was my client selected?”The employer may have an uncomfortable answer.“The system identified him.”That is not an explanation.The next question is obvious:“What did the system identify?”If the answer is a collection of thousands of data points processed through a proprietary model that the employer cannot reconstruct, the employer may have an extraordinarily sophisticated piece of technology and an extraordinarily weak justification for its decision.Traditional management decisions usually leave a trail that someone can explain. A manager observed something. A customer complained. A supervisor reviewed the employee’s work. An investigation uncovered particular conduct.AI can obscure that trail. It can identify a statistical pattern without establishing what caused it. An employee labelled “high risk” for misconduct has not necessarily committed misconduct. An applicant screened out by an algorithm has not necessarily been shown to be unsuitable.A prediction is not a fact. A correlation is not misconduct. A risk score is not proof. And a recommendation is not a finding.This advertisement has not loaded yet.This advertisement has not loaded yet, but your article continues below.The greater danger is what happens next. Once managers receive an AI-generated conclusion, they may start looking for evidence that confirms it rather than asking whether the conclusion is correct. That is an old management problem wearing new technology.There is another problem employers should consider before buying these systems. What happens when the vendor cannot explain the output either? Employers may assume that purchasing a sophisticated system means they have purchased sophisticated evidence. But they have not.Before deploying AI in a consequential employment process, an employer should understand what the system is designed to predict or identify, what information it uses, what limitations the vendor acknowledges and whether the employer can obtain enough information to investigate an adverse result.A black box may be acceptable for some low-stakes administrative functions. It is much harder to justify when someone’s job depends on its output.Employers should also consider what happens when the system changes. AI tools can be updated, retrained or replaced. An employer investigating a decision made with the help of AI months after the fact may be dealing with a system that no longer produces the same result from the same information.This advertisement has not loaded yet, but your article continues below.That makes record keeping particularly important. The employer should preserve the relevant output and enough surrounding information to establish what the system produced at the time, subject to applicable privacy and retention requirements.Before allowing AI to materially influence a consequential employment decision, employers should be able to answer five questions.First, what role did the AI play? Was it performing an administrative task, or did its output influence management’s decision?Second, what information did it use? The employer should know what information went into the system and preserve the relevant output for potential future litigation.Third, what does the AI actually establish? A score, prediction or recommendation should be treated as an investigative lead, not as proof of poor performance, dishonesty, misconduct or unsuitability.Fourth, what did the human decision maker independently verify? The employer should be able to identify the facts supporting the decision and anything that might undermine it.Fifth, can the decision be explained in ordinary language? The decision maker should be able to describe what the AI contributed, what was independently established and explain why they decided to act in the way they did.This advertisement has not loaded yet, but your article continues below.Those questions should also shape the employer’s dealings with the vendor. “Our algorithm is proprietary” may protect intellectual property, but it should not leave the employer unable to understand a consequential employment decision.The employer does not need the source code. It does need enough information to understand the system’s purpose, relevant inputs, limitations and output, so that it can determine whether a result makes sense in the particular case.AI may become an extraordinarily powerful management tool. But when the lawsuit arrives, the employer must still be able to explain what happened, what it knew and why it acted. Technology can assist management decisions. It cannot take responsibility for them.Howard Levitt is senior partner of Levitt LLP, leading his teams of labour lawyers in Ontario, Alberta and British Columbia. Howard has appeared in more Supreme Court employment law cases and provincial appeals than any lawyer in Canadian history. A bestselling author, he discusses current workplace issues on the podcast, At Work with Howard Levitt.This advertisement has not loaded yet.Notice for the Postmedia NetworkThis website uses cookies to personalize your content (including ads), and allows us to analyze our traffic. Read more about cookies here. By continuing to use our site, you agree to our Terms of Use and Privacy Policy.