by: Committee

These are a few modest stories some of us have shared about encounters with AI at our workplaces. Do you have your own story to tell? Contact us , and we’ll publish it in the next round!

An Engineer

AI usage was added as a section in promotion documents. Within 2 days a Junior developer presented a solution that did not need AI but had it. They cannot explain why it is needed over an alternate solution. Eventually they admitted it is because the company wants AI in solutions and they don’t think it is needed either.

An Engineer

I shared the big AI points of climate impact, surveillance, job threats, impact risks with some friends and asked if they wanted to help with protests or talking to others. A few said they were too busy. One said they didn’t want to risk their job. Three said they did not want to put their visa status at risk. If I lose my job, I have friends and family nearby to rely on if I cannot find another job soon. With AI being used to fuel layoffs, longterm unemployment seems likely. It also seems likely most hiring is being done to work on AI in this field. In this country we also have less social services to fallback on than ever. And if you are on a Visa getting deported to your country of origin is almost a best case now, with worst cases including deportation to the wrong country, imprisonment in a detention facility, or disappearing. I cannot argue from a place of privilege that they should risk their lives for this. But this added lever is being used: deportation, social service cuts, AI are all being used to scare workers and take away their security.

A Dataworker

A dataworker sits down and looks over the jobs available on the various platforms they work on. They only get paid for the moments they are actively in a job. If the job is not rejected. The hunt has to be fast to get the good paying work. FOMO (fear of missing out) keeps them at their computer for long hours, even when the work is not there, and earnings are not happening. They chose a job based on a few lines of text and a payment amount. The job requester could be lying about the hourly rate, they may not give good instructions, it’s all a gamble, every job, every day. The worker does not know what this work is going towards, and in many cases, not even the identity of the job requester. Is this facial recognition going to help ring cameras, or predatory military operations? Is identifying hate speech for moderation, or building a harmful chat model? The struggle between making money and keeping morals is ever-present, and seldom with all the necessary information.

When it’s all done, and the day’s goal has been met, they wonder how much of the work will be approved and when. Maybe they work a little longer for a buffer. Just in case. Just in case they don’t get paid, and just in case they don’t get paid for a month. But not too big of a buffer. Every unapproved job is a risk of rejection and harm to their account. Their livelihood could be taken away with no cause for recourse and no societal support when it’s gone. After all, they are not employees, they are not even called workers. They are participants, subjects, and contributors. Never workers, never protected, only exploited.

An Engineer

I recently conducted an interview for a role—the candidate was asked to talk about a project they’re proud of, and they talked about a GenerativeAI setup they did. It made the interview very difficult because the actual technical task was minimal, and hence I couldn’t really get a sense for the candidate’s abilities. Furthermore, it was obvious none of the possible downsides of the technology were addressed, there was no effort to watch for hallucinations, drift etc., it was just assumed if the information looked good for a few days they could put it into production and never revisit. Crazy on so many levels:

• The hype means the only acceptable projects to talk about now involve GenAI if you are looking for a job.
• People are not actually taking the same kind of care and perspective around GenAI projects that they would around their normal coding.