Market Frenzy Pressures AI Engineers to Burnout and Despair
You may have heard that we’re in the midst of an AI market frenzy. You might choose to call it an AI gold rush, or, if you’re less favorably disposed, an AI mania. Whatever you choose to call what’s happening today, I think we can all agree that we’re living through a manic, if not maniacal, period in the technology industry.
It’s a race for mindshare, market share, and ultimately market dominance. The early AI leaders today, in everything from models and services to underlying AI infrastructure, are sprinting to solidify and extend their advantages, while rivals clamber to gain ground before the market matures and consolidates.
Ultimately, huge sums of money, perhaps unprecedented riches, are at stake.
Pressure gains velocity as it descends from boardrooms, and from the executive suites along the salubrious corridors of mahogany row, down through the rank and file of organizations pursuing the glittering prize. If I may put it bluntly and in scatological terms, begging your forgiveness for the indelicate turn of phrase, shit rolls downhill. Business pressure mounts on the AI engineers mandated to transform vague but tantalizing promise into substantive reality or at least into something that represents a semblance of the popular conception of AI magic. The output doesn’t necessarily have to be effective and functional in a production environment, but it must possess enough gimcrack sizzle to dazzle prospective investors.
The Toxic Twins: Greed and Fear
There’s a lot on the line, bringing tensions to the fore. Greed and fear push and pull at maximum force. As pressure builds, those laboring to transform grandiose visions into something approximating reality feel the cumulative strain of impossible deadlines and capricious demands.
A feature article at CNBC, written by Hayden Field and published today, quotes several AI engineers from “Big Tech” companies (Amazon, Google, Microsoft, Apple), as well as those employed at other firms, attesting to the pressure and burnout that accompany development of ill-defined AI tools at breakneck speed. The objectives of these hasty development forays often involve appeasing investors by delivering superficially impressive demos rather than building functional products that provide real-world solutions for customers.
Nothing I read in that article surprised me. What that tells me is that the behavior we’re seeing now is not new, though the circumstances are different. Science and technology are artefacts of humans, but human behavior evolves at a more leisurely pace, providing a rich paradox to humanity’s material progress. To paraphrase an old chestnut, history might not repeat itself exactly, but it sure does rhyme.
There’s no question that the ever-quickening advance of technological progress during the past two centuries, and especially the impressive progress of information technology within the last 50 years, stands as an astonishing accomplishment. Unfortunately, people, including many tech luminaries lionized in the business press, remain afflicted with the same intellectual and ethical blind spots that hobbled the robber barons of the 19th century. As a result, the journey to technological advancement will remain, as ever, unnecessarily distressing and painful for a lot of folks assigned to do the grunt work.
What follows is an excerpt from the article:
Engineers and those with other roles in the field said an increasingly large part of their job was focused on satisfying investors and not falling behind the competition rather than solving actual problems for users. Some said they were switched over to AI teams to help support fast-paced rollouts without having adequate time to train or learn about AI, even if they are new to the technology.
A common feeling they described is burnout from immense pressure, long hours and mandates that are constantly changing. Many said their employers are looking past surveillance concerns, AI's effect on the climate and other potential harms, all in the name of speed. Some said they or their colleagues were looking for other jobs or switching out of AI departments, due to an untenable pace.
This is the dark underbelly of the generative AI gold rush. Tech companies are racing to build chatbots, agents and image generators, and they're spending billions of dollars training their own large language models to ensure their relevance in a market that's predicted to top $1 trillion in revenue within a decade.
It’s All About the Money
Of course, the investors, board members, and CEOs at these companies would earnestly tell you, especially when they’re being interviewed, that it’s not about the money. When they say that, dear reader, you can be assured that it’s irrefutably about the money. When this much money is at stake (just think of it, more than $1 trillion!), people behave strangely, afflicted by tunnel vision as they scramble up a slippery slope toward the AI treasure chest.
Ethical considerations aside (why start now?), is this really the best approach to achieve efficacious outcomes? Something akin to Gartner’s “hype cycle” recurs like a bad habit in the technology world, with a “peak of inflated expectations” ultimately followed by an occasionally protracted “trough of disillusionment.” True, genuinely useful and productive technologies ultimately emerge and prove themselves, but wouldn’t we save a lot of energy, time, and – yes, it must be conceded – money, if we had the discipline and maturity to skip the preceding delusions, madness, and nonsense that take us from the sugar high of irrational exuberance to the inevitable low of disappointment?
I know, I know. These are troubling and, in the end, perhaps rhetorical questions, and they lead to the glum precincts of introspection and rumination, locales seldom visited when punishing deadlines demand prompt and obedient execution.
Could it be, though, that the race might not go to the competitor that explodes from the starting blocks at full tilt, running exhaustingly in a zigzag pattern, but instead to the contestant that plots a carefully calibrated, considered, and direct course? We might never find out, but it seems a reasonable assumption.
Toward the end of the CNBC article, Ayodele Odubela, a data scientist and AI policy advisor, makes a perceptive observation:
"The biggest piece that’s missing is lacking the ability to work with domain experts on projects, and the ability to even evaluate them as stringently as they should be evaluated before release," Odubela said, regarding the current ethos in AI.
At a moment in technology when thoughtfulness is more important than ever, some of the leading companies appear to be doing the opposite.
"I think the major harm that comes is there's no time to think critically," Odubela said.
That quote offers a particularly savory ironic morsel. In the frenzied race to create unsurpassed artificial intelligence, leading market participants are eschewing the seemingly essential ingredient of critical thought. As such, they are more likely to produce artificial imprudence than artificial intelligence.