Can't find what you're looking for?
View all search results
I didn’t realize an artificial intelligence bot could have an opinion in this meeting. I knew it was attending, recording and transcribing the entire conversation, but I didn’t expect my supervisor would rather listen to the bot’s opinion than their own staff.
The bot in question was Zoom’s AI Companion, a recent favorite at an organization where I was interning. Embedded into online meetings, it could summarize discussions remarkably well. After the meeting, you could ask it questions and it would dig up answers based on the meeting transcript and any messages sent in the chat: “Give me a learning plan”; “the SWOT analysis”; “the weaknesses and strategies going forward”; “Evaluate the speakers.”
“Did you use an AI companion to write this proposal?” my supervisor asked me after the meeting.
I rarely find AI’s insights useful. Perhaps it’s my prompting technique, perhaps it’s my ambivalence toward AI. The prediction machine certainly had its novel moment back in 2023, but that quickly wore off. Hallucinations mean it isn’t always as “intelligent” as it’s marketed to be.
So for that assignment, I relied on my own intuition, just as I always had before 2023. I told my supervisor straight up: “No.”
“Oh, could you redo the proposal and use the AI Companion?”
I was surprised. I hadn’t expected a response that placed AI on such a high pedestal. I left the room wondering, is my work now secondary to AI?
A new work output standard?
At this point, it would take more effort to escape generative AI than to embrace it. Its presence has become so pervasive that even if we’re not actively participating in it, we’re probably already using its features. It summarizes our emails, sits in our Google Drive for smart search, generates images in Canva and powers our photo filters, among many other things.
These tools come in various forms with different functionalities, but they all seem to promise us one thing: that we can be better workers.
It’s no surprise, then, that companies are enthusiastic about adopting AI.
A Weekender survey of Indonesian workers found that more than 90 percent of respondents said their workplaces were encouraging the use of AI or building their own internal AI infrastructure.
They said AI was used predominantly as a “brainstorming partner” or “grammar consultant”, to “write or edit drafts”, “prepare financial reports” and “create analysis frameworks”, among other things.
But AI isn’t only changing our workflows. It’s also changing who is listened to at work.
Though half of the respondents said there had been no changes in their managers’ expectations, it’s worth noting that 44.7 percent said: “My manager expects me to consult AI first before submitting any work.”
“My manager scolds me if I don’t use AI,” wrote one respondent.
“Every solution starts and ends with AI. I don’t like how I have to consult AI about things I have already done,” wrote another.
Some chose to consult AI before their own managers. Others used AI as a mediator to reach consensus.
AI has gone from being a tool workers might use to becoming part of the standard against which workers are evaluated.
I remember what a supervisor once told me: “AI understands what I had in mind better than my own staff.”
Does that mean human judgment is being judged differently now? Are we creating some strange new cyborg standard for work output that requires both human and AI inputs?
What does it mean to show our shortcomings in the age of AI? What does it mean to flaunt our imperfections? Is there still a place for human clunkiness and strange digressions? Is there room for failures, disappointment and genuine learning opportunities in the workplace?
How much does it really help?
I get it. Businesses want more efficiency and clients want more polished work. Most workers understand how AI benefits them, too.
Rania, who works at a multinational tech company in Jakarta, says that AI has expanded what she can do.
“For projects that are small enough, I no longer need other teams to help me,” she says.
“I’m not just a marketing girl, I’m also a data wizard. I can build stuff, even though I’m not the one doing the coding.”
Though her overall work scope has expanded due to AI and she has concerns over the technology’s environmental impact, “I’m inclined to say that AI’s a good thing,” she says.
Similarly, 66.7 percent of survey respondents said, “Using AI has greatly improved the output of my work.”
When asked how, 64.9 percent said it “saves time”, 37.8 percent said it “helps me make better decisions informed by data” and 35.1 percent said it “allows me to be more creative”.
The remaining 21.6 percent said: “It helps, but it doesn’t significantly improve my output.”
Nadine falls into the last category. Even though she doesn’t think generative AI necessarily improves her output, the graphic designer finds that her clients and supervisors are expecting more from her. Working in advertising, her main job involves pitching ideas to clients.
“We used to start with raw ideas and sketches,” she says. “But now our pitches need to look real, as if we’re already producing the actual outputs.”
Nadine uses image generators to do this. Though it’s helpful to be able to illustrate abstract ideas, the resulting images still require significant editing by actual people, so she doubts that it saves any time at all.
AI may allow us to create outputs not possible within human timeframes, but these productivity gains are eventually eliminated because they become new productivity expectations.
And these gains come at a cost.
To struggle is human
A huge part of any creative work, be it corporate strategy or copywriting, is the struggle.
It’s the brick walls and dead ends we encounter in the process, the self-contradictions that refuse to be ironed out, the meandering, the getting lost, the returning with a messy but intricate map of our ruminations. None of these have shortcuts, and sometimes they won’t conform to a certain timeline.
A lot of the time, it’s also about making room for failure, something that most businesses despise and can now conveniently avoid by using the safer options generated by AI.
Researchers are already warning of the consequences of losing this friction in our learning process. Our outputs may appear improved, but excessive offloading of tasks to technology means we’re losing opportunities to learn or build the underlying skills.
Solving a problem is often what teaches us something, and if AI solves it before we even feel the friction, there's nothing to learn from. We're now willing to filter our imperfections with AI. It’s a double-edged sword.
One respondent said that with AI, “People are more confident doing written communication in English because at least they know that their writing is not grammatically wrong.”
But some feel the technology is undermining their confidence.
When I asked a teammate why he sent me AI-generated feedback, he simply said: “I’m scared to make a mistake. I’m not good at assessing other people’s work.”
Embracing imperfections
Maybe the answer isn’t to reject AI completely but to become more deliberate about what we choose to outsource to it.
As polished, AI-assisted work becomes the norm, some are beginning to see the value in the imperfections that machines are designed to smooth over.
“I’d rather hear broken English,” says Bella, an English studies graduate and philosophically inclined artist.
They are talking about local activists who choose the perfect English generated by chatbots to write social media captions rather than hiring someone.
It’s a little ironic, Bella says. Activists often promote fair compensation for workers yet still choose a machine, though perhaps not entirely through their own fault, to ensure that the capitalist metrics of clicks and followers are met.
Is it now time to value broken English simply because it’s human?
As much as AI can be an excellent assistant in terms of efficiency, for most people it offers temporary, sometimes superficial relief from the difficult work that helps us learn and grow. The more we outsource that work to AI, the less opportunity we have to develop the judgment that comes from doing it ourselves.
As we train AI to our liking, AI is training us as well. And there’s a chance it’s training us to be dull.
And it is already shaking something more fundamental, prompting us to question what it means to work, even what it means to be human.
It’s perhaps a little far-fetched to say that AI will replace human work entirely. But settling on any verdict now is perhaps premature. This uneasy middle ground we’re in could be a good vantage point, no matter how dim, for figuring out how we can best use AI to our advantage.
Perhaps for every time we improve our “AI skills”, we should also spend the same amount of time, if not more, refining our individual sensibilities: embracing the quirks, shortcomings and flaws and taking delight in being imperfect humans.
