ChatGPT Has A Serious Problem
- Thread starter Phil D.
- Start date
- Status
Polanus1561
Puritan Board Junior
Building ChatGPT’s AI content filters devastated workers' mental health, according to new report
A new report from 'Time' magazine paints a grim picture of labor conditions involved in training AI to avoid harmful and graphic content.
www.popsci.com
OpenAI Used Kenyan Workers Making $2 an Hour to Filter Traumatic Content from ChatGPT
Despite their integral role in building ChatGPT, the workers faced grueling conditions and low pay.
www.vice.com
"Despite their integral role in building ChatGPT, the workers faced grueling conditions and low pay. One Kenyan worker who was responsible for reading and labeling text for OpenAI told TIME that “he suffered from recurring visions after reading a graphic description of a man having sex with a dog in the presence of a young child.” The workers took home wages between $1.32 and $2 an hour, based on seniority and performance. "
MyCrows
Puritan Board Sophomore
Ummmm.This may be worse.Building ChatGPT’s AI content filters devastated workers' mental health, according to new report
A new report from 'Time' magazine paints a grim picture of labor conditions involved in training AI to avoid harmful and graphic content.
OpenAI Used Kenyan Workers Making $2 an Hour to Filter Traumatic Content from ChatGPT
Despite their integral role in building ChatGPT, the workers faced grueling conditions and low pay.
"Despite their integral role in building ChatGPT, the workers faced grueling conditions and low pay. One Kenyan worker who was responsible for reading and labeling text for OpenAI told TIME that “he suffered from recurring visions after reading a graphic description of a man having sex with a dog in the presence of a young child.” The workers took home wages between $1.32 and $2 an hour, based on seniority and performance. "
Wut?
MyCrows
Puritan Board Sophomore
"Time reported that content moderators at Sama who worked on projects for Meta became traumatized after viewing images and videos of executions, rape, and child abuse for $1.50 an hour."
To be fair, in the US we have a huge consumer base that actually pays to watch/ participate in simulated executions for hours on end.
To be fair, in the US we have a huge consumer base that actually pays to watch/ participate in simulated executions for hours on end.
arapahoepark
Puritan Board Professor
I'm having trouble following...are facebook users creating this content and its having to be monitored by the 3rd world?
MyCrows
Puritan Board Sophomore
It seems to me that the data companies were using intentionally created graphic media for the analysts to filter, repeatedly.
I was referring to our culture's obsession with murder/sexual content being commercialized. Think Grand Theft Auto, the legalized sex entertainment industry.
The 3rd world guys are being exposed to it trying to feed their families. We're taking away from our own families to revel in it.
I was referring to our culture's obsession with murder/sexual content being commercialized. Think Grand Theft Auto, the legalized sex entertainment industry.
The 3rd world guys are being exposed to it trying to feed their families. We're taking away from our own families to revel in it.
Ryan&Amber2013
Puritan Board Senior
So what does artificial intelligence have to do with graphic content? I can't figure out how to make that connection. Thanks!
After watching a video by Matt Whitman looking to see what chat gpt's theology was, I tried to do something similar. It seems that, when you ask in a way that it gives a clear answer, you get the loudest positions (not necessarily majority). In the example of theology, I got mainstream Evangelical answers on most issues, but if you ask the right questions, it will tell you children are full members of the church and that there are seven sacraments.
MyCrows
Puritan Board Sophomore
Honestly I don't really know what they are doing. I thought using regular expressions(advanced search functions) and keywords/phrases would be adequate to detect harmful content. I have no idea why they would need people to continually review graphic content that is that specific.
Polanus1561
Puritan Board Junior
Don’t know if this is linked. But there is a market for A.I chat bots that serve as companions with explicit language
YT&Reformed
Puritan Board Freshman
A lot of people believe ChatGPT is going to "take over" this and that, but I'm of the opinion that it will not. The system has several flaws that will prevent it from being adopted in any serious capacity in its current iteration.
1. It doesn't have access to the current internet, just a snapshot of the internet, so it cannot respond in real time
2. It can be modified by accessing sub-routines which allow it to "role play" as a bot with different rule-sets, this will discourage companies from adopting it
3. It has a clear bias coded into it, which companies on all sides of the political spectrum recognize as a critical issue
4. Just like any piece of tech, it is coded by people, and since people cannot create intelligence, it will likely suffer the flaws of whoever created it
That being said, it serves as a prototype of what is likely going to come. It will likely be the case that a more mature version of this type of tech will start to replace antiquated systems and you will see AI being integrated with your phone and other smart devices.
1. It doesn't have access to the current internet, just a snapshot of the internet, so it cannot respond in real time
2. It can be modified by accessing sub-routines which allow it to "role play" as a bot with different rule-sets, this will discourage companies from adopting it
3. It has a clear bias coded into it, which companies on all sides of the political spectrum recognize as a critical issue
4. Just like any piece of tech, it is coded by people, and since people cannot create intelligence, it will likely suffer the flaws of whoever created it
That being said, it serves as a prototype of what is likely going to come. It will likely be the case that a more mature version of this type of tech will start to replace antiquated systems and you will see AI being integrated with your phone and other smart devices.
alexanderjames
Puritan Board Sophomore
Hello friend of @danekristjan
I recall enjoying both your library tour videos a couple of years ago!
YT&Reformed
Puritan Board Freshman
Hello! I am new to Puritanboard, thanks for the warm welcome.
arapahoepark
Puritan Board Professor
Phil D.
Puritan Board Senior
News release March 15, 2023:
OpenAI has released GPT-4, the latest version of its hugely popular artificial intelligence chatbot ChatGPT.
...OpenAI said it had spent six months on safety features for GPT-4, and had trained it on human feedback. However it warned that it may still be prone to sharing disinformation.
...Like its predecessors, OpenAI has warned that GPT-4 is still not fully reliable and may "hallucinate" - a phenomenon where AI invents facts or makes reasoning errors.
OpenAI has released GPT-4, the latest version of its hugely popular artificial intelligence chatbot ChatGPT.
...OpenAI said it had spent six months on safety features for GPT-4, and had trained it on human feedback. However it warned that it may still be prone to sharing disinformation.
...Like its predecessors, OpenAI has warned that GPT-4 is still not fully reliable and may "hallucinate" - a phenomenon where AI invents facts or makes reasoning errors.
arapahoepark
Puritan Board Professor
All we need is a hippie Skynet.
MyCrows
Puritan Board Sophomore
That's just.... like.... your opinion, man.
Polanus1561
Puritan Board Junior
AI wants to be internet theologians?
Logan
Puritan Board Graduate
So...indistinguishable from a human? Does this mean it has passed the Turing Test?!
Phil D.
Puritan Board Senior
Haha.. Must admit I had to look that one up!
Phil D.
Puritan Board Senior
I'm not quite so optimistic. There are plenty of small businesses and mega corporations alike that have demonstrated an eagerness to jump on any opportunity to support leftist social engineering.
I recently read an article outlining how the U.S. military is hoping to soon start applying OAI's technology in such areas as intelligence analysis and automated targeting for weapons. ...Yikes..!
Last edited:
D
Deleted member 13126
Guest
The models have to be trained on large datasets. There's no such thing as 'unbiased'. For more information, Watanabe's ugly duckling theorem may be helpful. You have to give it enough right answers to a question that it can then guess an answer to a similar question with increasingly greater degrees of confidence. This is one of the reasons generative AI is very confidently wrong, and it takes someone who knows the right answer to know that an answer is wrong. So this is why third world workforces are being hired to help classify things for big companies like OpenAI.
Last edited by a moderator:
arapahoepark
Puritan Board Professor
I mean anything can happen...but this gave me quite a chuckle. The English taken over by Scotland? Hahaha. Not to mention the Danes are now Yugoslavia.
MyCrows
Puritan Board Sophomore
I see Robmania, but where's Miketopia?
BertMulder
Puritan Board Junior
Actually, the English have had a Scottish king since James the first...
- Status
Similar threads
- Deleted member 13126
- Computers & Technology
