The "Modern Day Slaves" Of The AI Tech World

null

Technology

Workforce

Social Media

Points Paras

Click

Use

to move to a smaller summary and to move to a larger one

The Reality of Artificial Intelligence and Hidden Human Labor

Behind the scenes of technological advancements, humans play a crucial role in inventing the future.
Companies like Google, Facebook, Amazon, and Uber rely on an invisible workforce to keep their applications running smoothly.
This workforce is often underpaid and disposable, earning as little as $5 an hour or even 10 cents an hour.
Some workers are hired to clean up social networks, but they must hide their involvement in these projects.
Going undercover as a web cleaner for Facebook, a content moderator reveals the disturbing things they witness and the impact it has on them.
The digital economy's best-kept secret lies in the factory of the future, where artificial intelligence is developed and refined.
Figures like Lucas Bald, founder of Figure8, specialize in AI and primarily work with Google.
Despite the glamorous image of tech companies, the reality behind the scenes involves humans doing the critical work.
These workers are vital to the success of companies like Twitter, American Express, and Tesco.

Figure8: How Humans Are Integral to Training AI Technology

Figure8, a company that specializes in training AI technology, believes in the importance of a loop between humans and machine learning.
The loop consists of humans doing the labeling and machine learning learning from the humans, creating a continuous feedback process.
For example, in developing self-driving cars, reliable systems need to be built to identify pedestrians, which is done through labeling photos.
Figure8 employs real people, referred to as contractors, to label photos and identify pedestrians, paying them for their work.
Contractors on Figure8's network can log in and work on these tasks from anywhere in the world.
Around 100,000 people consistently work on Figure8 tasks, with millions of others occasionally joining in.
Contractors are provided with a variety of tasks to complete, such as training search engines, and are paid for their work.

The Reality of Working as a "Ghost Worker" on Figure 8

Ghost workers are individuals who perform small tasks on platforms like Figure 8 to train search algorithms and AI systems.
These workers are not considered employees and do not have contracts or guaranteed wages.
The pay for these tasks is extremely low, with some workers earning less than one cent for every 10 answers.
Ghost workers can spend hours on these tasks and still earn less than a livable wage.
Figure 8 operates based on the supply and demand rule, where workers are paid based on the tasks available.
Ghost workers play a significant role in training AI systems that may eventually replace human labor.
Some ghost workers, like Jared, have other jobs but use Figure 8 as a way to earn extra cash.
The difficulty of earning a living through Figure 8 is evident, as even a team working for hours can earn only a few cents.
Don Carbone, a single mother, is an example of a full-time ghost worker who relies on Figure 8 for income.

Challenges and Controversies in the Figure8 Microtask Platform

Figure8 offers more work to experienced workers, but it may not always be more interesting.
Some tasks, like putting bounding boxes around people, may not be appealing to workers.
Finding viable and well-paid jobs on Figure8 can be a problem, especially in low job market areas.
Payment on Figure8 can vary greatly, from $5 per hour on good days to as low as 10 cents per hour on bad days.
Don, a worker on Figure8, prefers to work from home to take care of her daughter with autism.
Don receives government aid but signed up for Figure8 to earn additional income.
Figure8 pays a fixed rate, and refusing work means someone else will take it.
Lucas Bwal, the founder of Figure8, takes advantage of an unlimited supply of global ghost workers.
Clickworker and Amazon Mechanical Turk are other platforms offering similar underpaid online tasks.
Micro work is a growing concern for the International Labor Organization (ILO) due to low wages and lack of worker rights.

Lack of Concern for Crowdsourcing Workers

The average revenue per hour of contributors is highly variable and dependent on the task.
It is difficult to track time and ensure consistent pay due to the nature of online work.
The International Labor Organization (ILO) reported that crowdsourcing workers earn an average of $33-$31 per hour.
The company's payment policies may allow for payment below the American minimum wage.
The legality of paying below minimum wage is uncertain.
The discussion about crowdsourcing makes the interviewee uncomfortable and they would rather focus on AI.
The interviewee abruptly ends the conversation, leaving the head of PR to answer questions.
The head of PR also avoids discussing how the company treats its contractors.
The interviewee only wants to talk about artificial intelligence and not human labor.
The lack of concern for the well-being of crowdsourcing workers is attributed to the privileged background of Silicon Valley elites.

The Invisible Workforce Behind Social Media Platforms

There is a hidden workforce responsible for cleaning up and managing content on social media platforms.
These workers, referred to as "social media cleaners," are often not acknowledged or mentioned in the presentations of Silicon Valley CEOs.
Content reviewers, employed by subcontractors like Majorel, Cognizant, or Accenture, are responsible for sorting and managing content on platforms like Facebook.
The linguistic nuances of determining hate speech and inappropriate content still require human judgment, as AI tools are not yet advanced enough to accurately flag such content.
Facebook has approximately 15,000 content reviewers spread across various countries, including Ireland, Portugal, the Philippines, and the US.
Workers in this role are often underpaid, with monthly salaries barely above the minimum wage in their respective countries.
The job contracts for content reviewers rarely mention the name of the social media platform they are working for, even when directly asked.

Confidentiality and Training at Accenture for a Secret Project

Accenture refuses to disclose the name of the client for a project, even to new employees.
New employees attending a welcome meeting are informed that the client's identity is top secret and cannot be mentioned.
The project involves working for Facebook under a code name and confidentiality clauses.
Employees are not allowed to reveal their work for the project or mention the client's name to anyone outside Accenture.
Social media clean-up and confidentiality are emphasized, with potential legal consequences for breaking agreements.
Training for the project involves a three-week program on moderating Facebook content, including strict rules for decision-making.
Moderators must consider various factors, such as the impact on individuals and the context of the content, before making a decision.
Examples are given for different categories like nudity and dark humor to illustrate the complexity of the decision-making process.
Even artistic photographs with minor nudity can be challenging to handle, requiring careful interpretation of Facebook's rules.

Challenges and Impact of Content Moderation on Facebook

Content reviewers like Pedro and Greg face difficulties in their role of enforcing Facebook's rules without questioning them.
The job can become monotonous and robotic, with reviewers constantly processing large quantities of images and videos.
The exposure to disturbing and violent content can have a lasting impact on the mental well-being of moderators.
Moderators may not receive adequate preparation or support to handle the emotional toll of their work.
Some moderators may experience intrusive thoughts and difficulty separating their work from real-life situations.
Trainers and psychologists are available to provide support and help moderators cope with the traumatic content they encounter.
The psychological impact of the job can extend beyond the workplace, affecting how moderators perceive and interact with the world outside of their work.

The Invisible Labor Behind Social Media Moderation

People who work as social media moderators, often referred to as "content moderators", may experience post-traumatic stress disorder (PTSD) due to the disturbing and violent content they are exposed to every day.
Despite the negative impact, many moderators continue to work in this field as it is a job that provides a sense of security and a steady income.
Researchers estimate that over 30 people started working in moderation, but now only a handful are left due to the emotional toll of the job.
Moderators often feel a sense of guilt and responsibility for not being able to prevent violent or disturbing content from being posted online.
Companies like Facebook have confidentiality clauses that prevent moderators from speaking about their experiences, which can make it difficult for them to seek help.
The well-being of moderators is often prioritized last, and they are often underpaid and overworked.
The cost of moderation is becoming a significant economic burden for companies, leading them to outsource the work to low-wage workers or subcontractors to reduce costs.

Concerns and Lawsuits Surrounding Content Moderation on Facebook

Content moderators for Facebook experience a moral and emotional distance from the company.
Despite precautions, several former moderators filed lawsuits claiming post-traumatic stress disorder.
Facebook and its partners claim to prioritize the wellbeing of content moderators but refused interviews.

The Invisible Workforce behind Technological Advancements

Companies like Google, Facebook, Amazon, and Uber rely on an underpaid and disposable workforce to keep their applications running smoothly.
Workers are hired to clean up social networks and hide their involvement in these projects.
Figures like Lucas Bald, founder of Figure8, specialize in AI and primarily work with Google.
Workers are vital to the success of companies like Twitter, American Express, and Tesco.
Figure8 believes in the importance of a loop between humans and machine learning, where humans label and machine learning learns from humans.
Contractors on Figure8's network complete a variety of tasks such as training search engines and are paid for their work.
Ghost workers perform small tasks on platforms like Figure8 to train AI systems and are not considered employees.
Ghost workers earn extremely low pay, with some earning less than one cent for every 10 answers.
Figure8 operates based on the supply and demand rule, with workers paid based on tasks available.
Ghost workers play a significant role in training AI systems that may replace human labor.
Finding well-paid jobs on Figure8 can be a problem, especially in low job market areas.
Payment on Figure8 can vary greatly, from $5 per hour on good days to as low as 10 cents per hour on bad days.
Crowdsourcing platforms like Clickworker and Amazon Mechanical Turk also offer underpaid online tasks.
Micro work is a growing concern for the International Labor Organization (ILO) due to low wages and lack of worker rights.
The average revenue per hour for crowdsourcing workers is highly variable and dependent on the task.

The Hidden World of Social Media Cleaners and the Emotional Toll of Content Moderation

Social media cleaners are responsible for managing and cleaning up content on platforms like Facebook.
Content reviewers employed by subcontractors like Majorel, Cognizant, or Accenture sort and manage content on social media platforms.
Human judgment is still required to determine hate speech and inappropriate content, as AI tools are not advanced enough.
Facebook has approximately 15,000 content reviewers spread across various countries.
Content reviewers are often underpaid, with monthly salaries barely above the minimum wage in their respective countries.
Job contracts rarely mention the name of the social media platform they are working for.
Accenture refuses to disclose the name of the client, even to new employees.
Training for content reviewers involves a three-week program on moderating Facebook content and strict rules for decision-making.
The exposure to disturbing and violent content can have a lasting impact on the mental well-being of content moderators.
Moderators may not receive adequate preparation or support to handle the emotional toll of their work.
The psychological impact of the job can extend beyond the workplace, affecting how moderators perceive and interact with the world.
Content moderators may experience post-traumatic stress disorder (PTSD) due to their exposure to disturbing content.