A knife so sharp you don’t feel it cut

There’s an ongoing and heated debate about whether generative AI — like the technology that powers OpenAI’s ChatGPT — is either an overblown parlor trick, or a an existential threat to humanity that’s already raging out of control. The truth is of course some more nuanced point between these two extremes, but no matter what side you’re on, you should realize it’s already powerful enough to reshape industries and provoke another transformational shift in the broader working world.

It’s an odd feeling being employed in one of those industries that seems likely to be most impacted by the changes wrought by generative AI. Overall, it’s becoming clear that the fields which will feel it most will be white-collar and knowledge workers: It turns out it’s much harder to program a real-world robot to be able to interact with and manipulate its surroundings than it is to create a virtual sophist or a bot that groks the secret power language of code.

Some days, I honestly feel like how I imagine the dinosaurs would watching that extinction event meteor burning through the atmosphere: It feels momentous, but also too big to comprehend. At other times, there’s an awareness and a sense of urgency around efforts to anticipate and incorporate generative AI into our workflow and products. There’s a crackle and an energy towards action, but it’s also a bit flailing, a tad shapeless and carries just a whiff of desperation.

A few things are becoming clear regarding where and how generative AI can already excel, and genuinely replace a human equivalent. It’s exceedingly good at replicating the bland, toothless communication style prized in internal/external missives by large corporates whose primary motivating factor in most of those messages is avoiding offense. Basically, it’s perfectly fluent in CorpSpeak™ and pretty decent at bland and uninspired marketing copy — which probably makes up a significant percentage of the demand for marketing copy in terms of volume.

Like with the knife analogy in the headline, these scattered use cases feel innocuous at first — time-saving and productivity-enhancing helpers in a few instances. Who doesn’t want to offload that high-level executive summary, or exsanguinate their company-wide email to ensure it gets across the basic point in a way that’s guaranteed to never approach even a hint of offense? And who hasn’t dreaded having to write a brief description of an online seminar, or a 200 word bio for a speaking engagement.

It doesn’t feel deep — these types of tasks are the definition of shallow. On the coding side, it’s basically cobbling together readily available built examples to achieve a simulacrum of what you wanted that looks and feels so real it might as well be. Which is exactly what it’s doing on the language output side, too. But those who find comfort in this as a fixed depth limit are likely underestimating the pace of development on the one side, and the degree to which a facsimile of true depth is just as good as the real thing on the other. Also, it’s important to recognize that as shallow as they may feel, these examples are all things being done at massive scale currently by real, actual living breathing human beings who have little or no understanding that to a large extent, they may already be redundant.

There’s a recent example of an everyday job that’s relatively straightforward being slowly but surely subsumed by a technological solution that could stand as a good analogue for AI’s march: Self-checkout. When those self-checkout stations starting popping up at grocery stores and retail around a decade, customers and clerks alike looked at them as an odd, awkward and mildly humorous intrusion on their well-established space: Fast-forward to today, and they’ve but checkout staff to a tiny fraction of their former size at many medium and large retailers.

It’s still possible the cut is mostly superficial — but it’s also possible the blood is welling its way to the surface and simply hasn’t yet broken through. Plenty of other world-changing innovations seemed like toys or diversions at first, and were received as such by the people they ultimately impacted most, either out of genuine ignorance, or out of a reluctance to look an existential threat in the eye. That more or less applies to the printing press, the automobile, the computer, the smartphone and the internet. Which still leaves the question of what to do about it — or whether the time for that question is already passed.

Source link