Oh, there's better to do, put "'down" all the lying mainstream media sites (or almost all of them, because they generate an incalculable mass of traffic!
(If you know what I mean...)
We tested ChatGPT: ask your questions here without an account
-
- Moderator
- posts: 79364
- Registration: 10/02/03, 14:06
- Location: Greenhouse planet
- x 11060
Re: We tested ChatGPT: ask your questions here without an account
The first thing to delete would still be the RS starting with facebook, twitter and instagram... to gain human cognitive energy!
Although more energy is consumed by netflix and youtube...but this is to relax his intelligence!
Although more energy is consumed by netflix and youtube...but this is to relax his intelligence!
0 x
Do a image search or an text search - Netiquette of forum
-
- Moderator
- posts: 79364
- Registration: 10/02/03, 14:06
- Location: Greenhouse planet
- x 11060
Re: We tested ChatGPT: ask your questions here without an account
Not so artificial, GPT!!
https://www.novethic.fr/actualite/socia ... 51309.html
ChatGPT: behind the craze, exploited Kenyan workers
ChatGPT never ceases to be talked about. If the prowess of this new conversational agent using artificial intelligence concentrates most of the public's attention, behind the scenes of its design are less brilliant. Far from the Californian offices where OpenAI, the start-up behind the tool, is located, a low-paid workforce with difficult working conditions works in the shadows to limit the slippages of the chatbot, reveals a survey of Time.
"Super powerful", "bluffing", a "revolution"... On social networks, there is no shortage of superlatives to describe ChatGPT. This conversational agent, which has become a truly global phenomenon, regularly ranks first in Twitter trends with 64 messages the day after it went online and nearly 000 million tweets recorded in less than two months, according to Visibrain, a monitoring platform social networks. The principle: ask a question or write an order, to which an artificial intelligence responds in a precise and articulate way. Launched in November 2,5 by the Californian start-up OpenAI, the chatbot has gathered more than a million users in less than a week according to its president Greg Brockman. However, in the shadow of this dazzling success are hidden employees with trying working conditions.
This is the conclusion of an investigation revealed by Time on January 18. On the other side of the planet, Kenyan moderators reviewed thousands of violent texts over several months in order to develop the chatbot's security mechanism. During its learning phase, the artificial intelligence used by the latter has indeed used a large database drawn without filter on the internet. The result is prejudice, hateful or explicit remarks that OpenAI does not wish to confront its users. To prevent ChatGPT from echoing it and learning to detect this content, the start-up used a subcontractor company based in Kenya from November 2021.
"It was torture"
For less than two dollars an hour, the employees had the mission of labeling "tens of thousands of snippets of text" inappropriate, explains Time. Painstaking work that left the workers traumatized. “It was torture”, testifies one of them to the magazine. Several employees said they had to read between 150 and 250 passages of text every nine hours, some describing scenes of paedophilia, murder, torture or even incest. Accompaniment by "well-being" advisers was offered to them, but these sessions were rare and not very effective according to the employees.
A situation refuted by Sama, the subcontractor in charge of the service, which nevertheless decided to hastily terminate its contract with OpenAI four months later. Present in Kenya, India and Uganda, the company specializes in data labeling for tech giants, such as Google and Microsoft. Although Sama claims to provide "dignified work to people who need it", this is not her first scandal. The subcontractor had already been at the center of a Time investigation in February 2022. The media then denounced the harsh working conditions of employees, again in Kenya, responsible for moderating and removing violent or prohibited content from the social network. Facebook.
(...)
https://www.novethic.fr/actualite/socia ... 51309.html
ChatGPT: behind the craze, exploited Kenyan workers
ChatGPT never ceases to be talked about. If the prowess of this new conversational agent using artificial intelligence concentrates most of the public's attention, behind the scenes of its design are less brilliant. Far from the Californian offices where OpenAI, the start-up behind the tool, is located, a low-paid workforce with difficult working conditions works in the shadows to limit the slippages of the chatbot, reveals a survey of Time.
"Super powerful", "bluffing", a "revolution"... On social networks, there is no shortage of superlatives to describe ChatGPT. This conversational agent, which has become a truly global phenomenon, regularly ranks first in Twitter trends with 64 messages the day after it went online and nearly 000 million tweets recorded in less than two months, according to Visibrain, a monitoring platform social networks. The principle: ask a question or write an order, to which an artificial intelligence responds in a precise and articulate way. Launched in November 2,5 by the Californian start-up OpenAI, the chatbot has gathered more than a million users in less than a week according to its president Greg Brockman. However, in the shadow of this dazzling success are hidden employees with trying working conditions.
This is the conclusion of an investigation revealed by Time on January 18. On the other side of the planet, Kenyan moderators reviewed thousands of violent texts over several months in order to develop the chatbot's security mechanism. During its learning phase, the artificial intelligence used by the latter has indeed used a large database drawn without filter on the internet. The result is prejudice, hateful or explicit remarks that OpenAI does not wish to confront its users. To prevent ChatGPT from echoing it and learning to detect this content, the start-up used a subcontractor company based in Kenya from November 2021.
"It was torture"
For less than two dollars an hour, the employees had the mission of labeling "tens of thousands of snippets of text" inappropriate, explains Time. Painstaking work that left the workers traumatized. “It was torture”, testifies one of them to the magazine. Several employees said they had to read between 150 and 250 passages of text every nine hours, some describing scenes of paedophilia, murder, torture or even incest. Accompaniment by "well-being" advisers was offered to them, but these sessions were rare and not very effective according to the employees.
A situation refuted by Sama, the subcontractor in charge of the service, which nevertheless decided to hastily terminate its contract with OpenAI four months later. Present in Kenya, India and Uganda, the company specializes in data labeling for tech giants, such as Google and Microsoft. Although Sama claims to provide "dignified work to people who need it", this is not her first scandal. The subcontractor had already been at the center of a Time investigation in February 2022. The media then denounced the harsh working conditions of employees, again in Kenya, responsible for moderating and removing violent or prohibited content from the social network. Facebook.
(...)
1 x
Do a image search or an text search - Netiquette of forum
-
- Moderator
- posts: 79364
- Registration: 10/02/03, 14:06
- Location: Greenhouse planet
- x 11060
Re: We tested ChatGPT: ask your questions here without an account
If not, who has tested DALL.E here?
From what I see, there is a DALL.E function already integrated into ChatGPT but each time I asked ChatGPT for an image I got 404...
Has anyone had better luck than me?
From what I see, there is a DALL.E function already integrated into ChatGPT but each time I asked ChatGPT for an image I got 404...
Has anyone had better luck than me?
0 x
Do a image search or an text search - Netiquette of forum
-
- Econologue expert
- posts: 13718
- Registration: 17/03/14, 23:42
- Location: picardie
- x 1525
- Contact :
Re: We tested ChatGPT: ask your questions here without an account
GPTZero would be a solution for educators to detect homework done using ChatGPT? not tried https://gptzero.me/
1 x
Re: We tested ChatGPT: ask your questions here without an account
I farted?! Nothing but wind...and what a stench!
0 x
"We make science with facts, like making a house with stones: but an accumulation of facts is no more a science than a pile of stones is a house" Henri Poincaré
-
- Moderator
- posts: 79364
- Registration: 10/02/03, 14:06
- Location: Greenhouse planet
- x 11060
Re: We tested ChatGPT: ask your questions here without an account
Janic wrote:I farted?! Nothing but wind...and what a stench!
yes, I was called?
0 x
-
- Moderator
- posts: 79364
- Registration: 10/02/03, 14:06
- Location: Greenhouse planet
- x 11060
- Obamot
- Econologue expert
- posts: 28725
- Registration: 22/08/09, 22:38
- Location: regio genevesis
- x 5538
Re: We tested ChatGPT: ask your questions here without an account
At home we can say that it is not "zero emissions"Janic wrote:izentrop wrote:GPTZero
I farted?! Nothing but wind...and what a stench!
1 x
-
- Similar topics
- Replies
- views
- Last message
-
- 3 Replies
- 1950 views
-
Last message by Janic
View the latest post
02/10/23, 08:23A subject posted in the forum : Science and Technology
-
- 13 Replies
- 7365 views
-
Last message by plasmanu
View the latest post
30/05/12, 20:52A subject posted in the forum : Science and Technology
Back to "Science and Technology"
Who is online ?
Users browsing this forum : No registered users and 131 guests