When it Comes to AI, Clear-Eyed Pragmatism Beats Slaphappy Optimism

After my return from vacation, I had a lot on my plate these last few days. The preceding sentence is my way of apologizing for not providing a post earlier this week.

Still, I’ll attempt to make amends, offering this post today and another tomorrow (on Nvidia GTC . . . for subscribers), as well as three posts next week.

Just because I didn’t write anything in the last few days doesn’t mean the period in question was uneventful. If anything, the frenzied pace of activity, in the technology industry and in the wider world beyond, has intensified, with change the only constant. We are all afflicted by the curse of living in interesting times.

How interesting are our times? Well, it’s impossible to count all the ways, but we can make a start by enumerating them one at a time. Actually, taking things one at a time seems the only coherent way to proceed. I could try stream of consciousness, but you wouldn’t thank me for it.

First up, then, is news from Entrepreneur Magazine regarding Morgan Stanley’s plans to reduce its 80,000-employee workforce later this month by 2,000 employees. The company’s financial advisers are reportedly being spared, but the cuts will otherwise affect divisions spanning Morgan Stanley’s financial empire.

A 2.5% reduction in staff might not seem extreme, but there’s an AI angle to this story that suggests more cuts are forthcoming, presuming that AI can deliver on its cost-cutting promise. The following excerpt from the article cites AI’s role in Morgan Stanley’s (and other banks’) plans to shed flesh-and-blood personnel:

Some employees impacted by the layoffs will be let go due to performance issues, while others will be cut because AI and automation have replaced their roles within the bank. A source told Bloomberg that the bank expects to make more job reductions due to AI in the coming years.

Morgan Stanley isn't the only major bank planning to cut roles due to AI. A Bloomberg Intelligence report released earlier this year surveyed chief information and technology officers at 93 major banks, including JPMorgan and Goldman Sachs, and found that executives expect to lay off an average of 3% of their workforce within the next three to five years as AI takes over tasks. That means up to 200,000 jobs on Wall Street are at risk of being cut due to automation.

Automating for Dollars

Apparently, AI tools are proliferating within Morgan Stanley. In the second half of 2023, the bank launched a knowledge-assistant AI tool designed to quickly find and present the company’s research to its financial advisers. In June 2024, Morgan Stanley introduced a different AI tool that takes notes and identifies action items during financial advisers’ video calls with clients.

Morgan Stanley’s executives, whose positions atop the company are not endangered by AI, are bullish on what the technology can accomplish. In fact, the company’s executives already credit AI, as a means of making the bank more efficient and productive, for helping Morgan Stanley achieve record revenue and profits.

Prediction is a tough proposition, especially these days. Nonetheless, we can see that major banks are counting on AI to help them cull headcount and save money by offloading human tasks to automated processes. Perhaps the banks will cut an average of 3% of their staff members, but it’s possible the percentage will be higher or lower than that percentage. We can take an educated guess at where the number might land, but we would be doing nothing more than speculating. The fact is, we don’t know precisely how AI will perform in the real world as it ascends the value curve of increasingly sophisticated tasks. Maybe AI will blow past even the most favorable expectations, but maybe not.

Before it can rolled out extensively, replacing a growing number of humans in the workforce, AI will have to demonstrate that it can perform work adequately — not necessarily better than the humans it supplants, but at least to an acceptable standard.

Remember what used to be said about Microsoft’s litmus test for its software — that it had to be “good enough” to dissuade customers from incurring the cost and trouble of defecting to a competing vendor’s software? Well, a “good enough” rule for AI might come into effect, whereby the technology must clear the hurdle of approximating, but not necessarily equaling or surpassing, the quality of the output and work currently (or previously) done by human employees, who not only demand salaries but also qualify for a wide range of benefits.

The thing about AI, which broadly belongs to the category of software-based automation, is that it incurs none of the costs that typically accrue to human staff members — no recurring salary, no paid holidays, no sick days, no time off during the day for breaks or lunches, and no medical insurance or other benefits. The banking sector isn’t the only industry that will find the allure of AI irresistible.

Hope isn’t Strategy

That said, AI’s appeal might be marred by cost-cutter’s myopia, especially if all industries aggressively adopt the technology simultaneously. AI’s proponents invariably tout the technology’s ability to create new jobs, and that might occur at some point; but most of the emphasis today among enterprise customers is on AI’s capacity to facilitate staff redundancies, which manifest on the bottom line as efficiency-driven cost savings.

One can reasonably hypothesize that AI’s job-creating capacity, if it exists, will not materialize immediately. The initial focus, in banking as in other industries, is on how AI can supplant human employees. The AI-enabled human jobs of some indeterminate future will lag behind the jobs sacrificed on the altar of AI efficiencies. If all industries slash 3% of their human workers and replace them with AI-based intelligent automation, the contraction in the consumer economy would be significant, especially if AI eliminates relatively high-paying jobs among white-collar professionals, including knowledge workers. Would a recession result from, or be aggravated by, a fear-driven contraction of consumer spending? It’s at least possible, especially in an era where social programs and government largesse seem to accrue only to the billionaire class.

Nobody in the AI vendor community wants to discuss the economic and societal implications of AI. Instead, they wax merrily about AI rainbows, lollypops, unicorns, and a refulgent world of perpetual happiness and splendor.

Of course, we’ll only have to deal with these economic and social problems if AI meets its promise to be good enough to replace human employees across a broad spectrum of professions and industries.

Maybe AI doesn’t get there. Perhaps we have nothing, or little, to worry about.

Nonetheless, I’ve always embraced the dictum that we should hope for the best but prepare for the worst. Hope, as we all know, is not a viable strategy. Perhaps our deliverance, in a world increasingly defined by automation, will result less from subscribing to blithe optimism and more from adherence to clear-eyed pragmatism.