The perfect storm for AI disruption
Summary
There are widespread concerns about how AI will disrupt employment and render many people jobless. I argue that jobs that involve routine, non-novel, and acceptably risky knowledge work will be prime candidates for such disruption.
(a version of this article first appeared in Reworked magazine)
Some months back, I was on a sabbatical from work. In my second life as a wildlife photographer, I spent numerous days and hours in the wilderness of Central India and Indonesian Borneo, searching for elusive creatures and wildlife moments.
For the uninitiated, let me tell you that wildlife photography is one of the most frustrating crafts. You spend hours in the wilderness battling heat, dust, humidity and insects in the hope of striking gold. Eight times out of 10, you fail. The two times you succeed, you end up with images that resemble someone else’s. The genuinely unique photos can take a few seconds to snag or a few years, depending on your luck, skill and perseverance.
Wildlife photography involves much toil for fleeting rewards.
While I cut myself off from work during the sabbatical, I continued to read all the AI news because, let’s face it, you can’t escape AI these days! One recurring theme caught my attention — the warning about how artificial general intelligence or sentient AI will render millions jobless.
It made me wonder why, with all the imagery that AI can produce, I was toiling away in the forest, chasing those elusive photographs. Why does one watch a Christopher Nolan film or admire Christina Mittermeier’s photography? What about David Remnick and Ravish Kumar’s journalism or Trevor Noah and Vir Das’s comedy? Would an AI-generated image be just as evocative for us as Steve McCurry’s iconic photo of the Afghan girl? There’s something to recognize about why we create and crave to see another human being’s original work. As Navneet Alang eloquently explained in The Guardian,
“To say that AI on its own will be able to produce art misunderstands why we turn to the art in the first place. We crave things made by humans because we care about what humans say and feel about their experience of being a person and a body in the world.”
Human vs AI - not just about the outputs
Where AI will affect employment
I predict our inherent desire to create things and admire other people’s work will not change, regardless of how “intelligent” the machines become. But I also think it’s naive to say that nothing will change or to believe that AI will not impact employment. From all indications, four criteria determine how likely AI is to disrupt a job.
Routine: Repetitive, process-driven jobs are prime AI disruption or integration candidates. We’ve been seeing these innovations even before the LLM gold rush. Whether it’s the one-second heuristic in redaction algorithms or the more complex, assistive applications such as robotic surgical arms, autopilot on an aircraft and adaptive cruise control, “intelligent” algorithms have been part of our lives for a while now. Going forward, I expect even more AI innovation and disruption in this space. For example, consider if some or all parts copy-editing could be a machine’s job. How long before technical assistants and personal assistants find themselves out of jobs?
Non-novel: Modern generative AI capabilities remix corpora of knowledge and art to create something that looks original. But most of these outputs build on an existing sense of aesthetics and design standards. Remixing existing content and basing the output on established patterns is enough for many jobs. Take UI design, for example. Every user interface needn’t be the next award-winning, innovative screen. You could use a tool like Galileo to generate most of your design ideas based on your project’s design standards. A tweak here, an adjustment there, and Bob’s your uncle! Many software teams may not need a designer on board anymore, especially if the job is to create simple but visually pleasing screens that do the job.
Knowledge work: It’s harder for AI to be your safari guide, surfing coach or rescuer on Mount Everest than to be a software developer. Sure, there’s more to software development than generating code, but at least in the foreseeable future, AI has more chance to disrupt pure knowledge work jobs than jobs with a physical, real-world component. The IT industry is already seeing these impacts. We’re already building research assistants to analyze preclinical data in pharmaceutical research. If a job can happen from a desk today, it’s likely to be AI-augmented or replaced by AI in the future.
Acceptably risky: This brings me to the final criterion for AI disruption. The notion of “acceptable risk” is often emotional and belies logic. The higher the stakes, the less likely we accept machine failure. Let’s assume that autonomous vehicles are safer than the ones humans drive. But when a self-driving car meets with an accident, there’s outrage everywhere. We don’t wait to examine the research when the stakes are as high as life or death. Illogical as it may seem to the car manufacturers who’ll show you the safety stats, will the research paper assuage the emotions of someone who lost a family member? This acceptable risk assessment depends on the domain and can be subjective and political. In high-risk domains, David Heinemeier Hansson’s assessment will hold for at least a few years.
“Being as good as a human isn’t good enough for a robot. They need to be computer good. That is, virtually perfect. That’s a tough bar to scale.”
The conditions in which AI could disrupt a job
So, if our jobs are in that sweet spot — routine, non-novel knowledge work that’s acceptably risky to replace with machine intelligence — we must find ways to build skills that take us out of that zone. Thankfully, corporations are often slow to adopt tech, which buys us some of the time we need.
A warning and a possibility
As I strike this note of seeming urgency, let me conclude with the premise I began with. No AI will replace our desire to create or people’s desire to see other human beings' creations. Aside from all the possible disruptions, I expect AI to fuel a new creator economy. AI is already a creative partner in this economy - much like the auto-focus tools inside my camera or the editing tools that bring my photos to life. As we step out of that acceptably risky, routine, non-novel knowledge work sweet spot, let’s consider how AI can be an ally for our creative side rather than an adversary. Wouldn’t that be a future of work worth looking forward to?