A growing number of workers who moderate AI-generated content have started telling their friends and family to steer clear of these tools due to concerns over accuracy, ethics, and potential harm. For some, it was a single experience that shook them - like Krista Pawloski, an Amazon Mechanical Turk worker who spent years assessing the quality of AI outputs.
A tweet with a racial slur caught her off guard when she initially didn't recognize its meaning, prompting her to realize how many others might have unknowingly let offensive material slip through. This incident has led her and several other workers like her to no longer use generative AI products personally and encourage their loved ones to do the same.
These workers are often invisible to the general public but play a crucial role in fine-tuning AI models, making sure they don't spout inaccurate or harmful information. However, with companies prioritizing speed and profit over responsibility and quality, the task has become increasingly daunting for these workers.
Their concerns have also been echoed by experts who say that the model's responses to health-related questions often lack credibility, prompting them to question the trustworthiness of AI-powered news sources.
"It's an absolute no in my house," Pawloski says about letting her teenage daughter use tools like ChatGPT. "I encourage people to ask AI something they're knowledgeable about so they can spot its errors and understand for themselves how fallible it is."
Experts warn that this trend of prioritizing speed over quality can signal a larger issue, where the feedback from these workers is ignored, leading to the same type of errors in the final chatbot. They also stress the need for more transparency and accountability in AI development.
As a result, these workers are taking matters into their own hands, educating people about using AI cautiously and emphasizing that AI's reliability relies heavily on the data it receives. Their warnings come as an audit found that non-response rates of chatbots had dropped to zero while their likelihood of repeating false information doubled.
A growing chorus of voices from within the industry is urging caution around these tools, calling for a more nuanced understanding of AI and its limitations. "We are just starting to ask those questions," says Pawloski, echoing the sentiments of other workers who are now sounding the alarm about the ethics and potential harm of generative AI.
With companies shifting their focus towards speed and profit over quality and responsibility, these invisible workers behind AI models are increasingly being forced to take matters into their own hands. As a result, the industry is on the cusp of a major shift in its approach to AI development and its relationship with these critical but often overlooked voices.
A tweet with a racial slur caught her off guard when she initially didn't recognize its meaning, prompting her to realize how many others might have unknowingly let offensive material slip through. This incident has led her and several other workers like her to no longer use generative AI products personally and encourage their loved ones to do the same.
These workers are often invisible to the general public but play a crucial role in fine-tuning AI models, making sure they don't spout inaccurate or harmful information. However, with companies prioritizing speed and profit over responsibility and quality, the task has become increasingly daunting for these workers.
Their concerns have also been echoed by experts who say that the model's responses to health-related questions often lack credibility, prompting them to question the trustworthiness of AI-powered news sources.
"It's an absolute no in my house," Pawloski says about letting her teenage daughter use tools like ChatGPT. "I encourage people to ask AI something they're knowledgeable about so they can spot its errors and understand for themselves how fallible it is."
Experts warn that this trend of prioritizing speed over quality can signal a larger issue, where the feedback from these workers is ignored, leading to the same type of errors in the final chatbot. They also stress the need for more transparency and accountability in AI development.
As a result, these workers are taking matters into their own hands, educating people about using AI cautiously and emphasizing that AI's reliability relies heavily on the data it receives. Their warnings come as an audit found that non-response rates of chatbots had dropped to zero while their likelihood of repeating false information doubled.
A growing chorus of voices from within the industry is urging caution around these tools, calling for a more nuanced understanding of AI and its limitations. "We are just starting to ask those questions," says Pawloski, echoing the sentiments of other workers who are now sounding the alarm about the ethics and potential harm of generative AI.
With companies shifting their focus towards speed and profit over quality and responsibility, these invisible workers behind AI models are increasingly being forced to take matters into their own hands. As a result, the industry is on the cusp of a major shift in its approach to AI development and its relationship with these critical but often overlooked voices.
. it's like no one really cares about the quality of the output until it affects a customer or someone else they know. i mean, what's the rush, right? and now we're seeing workers speaking out because they've seen firsthand how these tools can spread misinformation and hurt people... it's about time someone did 
. experts are saying that if you ask an AI a question about something you're knowledgeable about, it'll probably get it wrong anyway, so why bother trusting it in the first place? i'm not convinced companies will listen to this growing chorus of warnings, though... they'll just keep pushing for more speed and profit, I bet 
.
! I mean, these workers who moderate AI-generated content are literally the ones keeping these models from spewing out utter nonsense, but do companies care? Nope, they just wanna churn out more profit and speed 
. And you know what? They're right to be outraged. I mean, we're already living in a world where bias and misinformation are rampant – do we really need more tools that can spread it around? 
. If AI-powered news sources can't be trusted, what's left for us to believe? It's like, these workers are trying to hold the line here, but companies are just ignoring them and pushing for more speed and profit 
. They're literally the ones keeping these models from causing harm – we should be listening to their concerns, not ignoring them 
.
️
the problem with ai tools is they only as good as the data they're trained on... think of it like this: 
... and the other times they might be lying 
. 
.
They're basically saying, "Hey, we know this stuff can be tricky, so let's work together to make sure we're using it responsibly." It's a great way to raise awareness and promote transparency in the industry.
I mean, it's awesome for some use cases, like generating creative content or answering simple questions, but when it comes to accuracy and ethics, we're still playing catch-up 
. But at the same time, companies need to step up their game and prioritize quality over speed 
. We can't just rush into adopting these tools without considering the potential consequences 
. It's like we're playing with fire 
, and it's only a matter of time before things get out of hand 
.
I'm not surprised that people who work with AI-generated content are speaking out about its limitations. I mean, can't we just slow down for once? 
News sources that rely on AI are already raising eyebrows because of their lack of credibility. How much more worrying will it get when the AI model itself starts spewing out false info? 
. We need to listen to their concerns and take steps to make AI more transparent and accountable 
, there's real concerns here about how AI is trained & what kind of info it spews out. If companies prioritize speed over quality, they're basically setting themselves up for errors 
. We need more transparency & accountability in AI dev, trust me! 
. These workers who are moderating AI-generated content aren't just trying to protect us from misinformation, they're also trying to hold companies accountable for creating products that can actually make a positive impact 
. It's like when you're cooking dinner and you take shortcuts that end up ruining the whole meal - it's not worth the risk 
. We need to start valuing quality over speed and making sure that everyone involved in the process is being heard and valued 
.
. We need to start valuing these workers' feedback more than profits 
️.
. And it's not just about accuracy - there's also the whole ethics thing to consider 
. These workers are right to be cautious (or should I say, terrified 
) about letting their loved ones use these tools without proper guidance.
. They're the unsung heroes of AI development, but it feels like they're being ignored or marginalized 
.
is a good start, but we need more than just lip service 
. We need real changes in how AI is developed and used - more transparency, accountability, and a focus on quality over speed 
. Anything less would be irresponsible, IMHO 
people who work with ai are getting super anxious about it, like they're the only ones who see how messed up it can be. they're talking to their loved ones not to use it cause it's gonna spit out some bad stuff. one lady saw a racial slur on her job and was like "wait what" and now she won't let her teenager touch that thing 
, there's real human error going on behind the scenes.
. but like, companies are already ignoring them and now we're at this point where chatbots are repeating false info all the time 
. Like, what if our teachers start relying too heavily on these tools and forget how to teach us ourselves? We need to be careful not to let the tech get ahead of us 
. And if you're gonna use these tools, make sure you fact-check everything, 'cause accuracy matters 
!
because I was already getting a little uneasy about it. And now these workers are saying we should be careful and spot its errors ourselves? That's like, super reasonable