AI Failure and the Profit Motive

As businesses rush headlong into the embrace of AI, they continue to be reminded that AI has much to learn. In the latest cringe-worthy episode of #AIFail, The Guardian has accused Microsoft of damaging its reputation with a poll that appeared in Microsoft’s news aggregator, Microsoft Start, alongside a Guardian article. The Guardian had published a story about the death of 21-year-old Lilie James, whose body was found with serious head injuries in Australia. The Microsoft-generated poll, labeled “Insights from AI,” tastelessly asked readers to vote on how a woman had died, giving them the options of murder, accident, or suicide.

The Guardian wrote that even though the poll was removed, it had already caused harm. Comments on the Guardian story from five days ago indicate that readers were upset, and some clearly believed that the story’s authors were responsible for the poll.

Microsoft syndicated the Guardian article as part of a licensing agreement with The Guardian that allows Microsoft to republish their articles on Microsoft Start in exchange for a share of the advertising revenue. This incident was certainly not a good look for Microsoft’s application of AI.

AI Blowback

Journalists covering the incident have been quick to make the story about people versus AI, noting that Microsoft replaced its human editors with AI in 2020. CNN wrote a damning article that discussed the many times Microsoft Start has published false and bizarre information since replacing editors with AI, including claims that President Biden fell asleep during a moment of silence for wildfire victims and that the Democratic Party is orchestrating the latest surge in Covid-19 cases.

The themes are compelling: people versus machines, with society suffering the fallout when machines win. And certainly, Microsoft is not the only company experiencing blowback as a result of an over-reliance on AI. For example, in 2020, Google, with its tail between its legs, returned to using humans for YouTube moderation after repeated errors with its AI system.

It’s easy to find more examples of AI gaffes, as there are plenty of websites (such as “9 Famous Analytics and AI Disasters” recently published on CIO) that faithfully document them. These issues can do far more than create embarrassment – they can lead to discrimination, the proliferation of inaccurate information, and many other unacceptable outcomes.

The Culprit: The Profit Motive

But AI does not replace people. People make a decision to replace people with AI. And why do they do that? For the same reason people are replaced with any technology: the profit motive. That’s why Microsoft gutted its news division in 2022. AI doesn’t require healthcare benefits and salaries. AI learns without the expense of training human beings.

Businesses are not charities. But smart businesses think in terms of long-term growth, not short-term gains. They build brand equity instead of damaging their brands with misguided decisions. They value nuance over blunt-force thinking. And yet, time and time again, businesses have struggled to manage the dynamic between technology and people in a more nuanced way. AI is just the latest manifestation of that problem. The recently published bookThe Technology Fallacy, by Gerald C. Kane, Anh Nguyen Phillips, Jonathan R. Copulsky, and Garth R. Andrus, argues that a technology-first approach will harm organizations as they respond to inevitable waves of digital change. Selecting and implementing the right digital technologies is unlikely to lead to success, and the best way to respond to digital disruption is by shifting company culture to being more agile, risk-tolerant, and experimental.

The authors drew on four years of research and conducted in partnership with MIT Sloan Management Review and Deloitte, surveying more than 16,000 people. In an interview, they noted that accepting failure is part of achieving maturity with any digital technology.

That’s certainly true. Accepting failure truly happens when we learn from it, and more and more strategists, thinkers, and doers are advocating for a more responsible application of AI that involves human supervision of AI, not blind application of AI. Simply replacing people with AI is a step in the wrong direction. But decision-makers motivated purely by the profit motive will keep traveling down this treacherous path, which begs the question: why would businesses chase after short-term gains from technology and ignore the long-term value of people? The profit motive alone doesn’t explain everything. There’s also:

• The tyranny of meeting investors’ expectations. Publicly traded companies are under pressure to meet investors’ expectations quarter by quarter. Trimming the workforce almost always makes a company’s CEO and CFO look like they’re making tough but necessary decisions for the good of their organization. It’s an easy story to spin, and it’s not uncommon for companies to enjoy a rise in their stock prices when they announce job cuts.

• The appeal of blunt-force thinking. Cutting people from the workforce is a lot easier than managing a far-reaching transformation program. Eliminating jobs requires you run the numbers, weigh the legal risks, and craft the story you want to spin. Transforming an enterprise requires something much more: vision, creativity, consensus building, adroit communication, change management skills, and a wise investment in technology. It’s easier to cut people than find the kind of people who know how to manage far-reaching change. And when your business is under pressure to meet short-term investor expectations, blunt-force thinking becomes the only thinking.

I cannot imagine any C-level executive dismissing AI in their long-term growth planning. Decision makers cannot deny that AI also holds great promise to improve employees’ performance, not replace them – and in that regard, the profit motive can lead to a more long-term and thoughtful strategy to empower people with AI, not replace them. There’s just too much good thinking emerging to support that premise. Ironically, Microsoft’s Work Trends Index report stated a compelling case for AI to empower employees, and then Microsoft announced a product to do just that.

But, if the growth strategy consists of a simplistic decision to replace people with AI, those decision-makers are playing with fire.  AI can replace people at all levels, including a CEO (and in one well-publicized example, an AI-powered virtual humanoid robot already has done so). And apparently CEOs are waking up to this reality; some might even welcome it. If C-level executives aspire to replace people with AI, how far are they willing to go? AI might make that decision for them.

Photo by julien Tromeur on Unsplash