the success story is not one of Silicon Valley genius alone. In its quest to make ChatGPT less toxic, OpenAI used outsourced Kenyan laborers earning less than $2 per hour, a TIME investigation has found. https://time.com/6247678/openai-chatgpt-kenya-workers/

Exclusive: The $2 Per Hour Workers Who Made ChatGPT Safer

Todd Smith

DeepScribe, an healthtech startup providing AI-powered medical records transcribing service, is relying on 200 contractors to correct errors, current and former workers told WSJ. The platform claims to turn doctors' conversations with patients into medical records using AI. However, due to the nascent technology, AI often adds incorrect medical terminology and medical data to records. 
- The San Francisco, Calif.-based startup is using human contractors as a stop-gap solution to "catch errors that can be a product of nascent technology such as AI and give healthcare providers the confidence of knowing that their notes are reviewed by a training member of our team for accuracy."
- Using humans to catch and rectify errors also helps improve the accuracy of the AI model. 
- Co-founders Matthew Ko and Akilesh Bapu recognized that many AI algorithms are prone to suffer from "hallucinations." 
- Their AI model is currently able to transcribe nearly 80% of each record correctly.

https://www.wsj.com/articles/at-startup-that-says-its-ai-writes-medical-records-humans-do-a-lot-of-the-work-794be22e?mod=Searchresults_pos1&page=1

DeepScribe, an healthtech startup providing AI-powered medical records transcribing service, is relying on 200 contractors to correct errors, current and former workers told WSJ. The platform claims to turn doctors' conversations with patients into medical records using AI. However, due to the nascent technology, AI often adds incorrect medical terminology and medical data to records.  - The San Francisco, Calif.-based startup is using human contractors as a stop-gap solution to "catch errors that can be a product of nascent technology such as AI and give healthcare providers the confidence of knowing that their notes are reviewed by a training member of our team for accuracy." - Using humans to catch and rectify errors also helps improve the accuracy of the AI model.  - Co-founders Matthew Ko and Akilesh Bapu recognized that many AI algorithms are prone to suffer from "hallucinations."  - Their AI model is currently able to transcribe nearly 80% of each record correctly.  https://www.wsj.com/articles/at-startup-that-says-its-ai-writes-medical-records-humans-do-a-lot-of-the-work-794be22e?mod=Searchresults_pos1&page=1

DeepScribe relies on a team of people to review tech-generated reports, listen to medical conversations and revise the records.

WSJ News Exclusive | At Startup That Says Its AI Writes Medical Records, Humans Do a Lot of the Work

Sam Mendoza

Annotation remains a foundational part of making AI, but there is often a sense among engineers that it’s a passing, inconvenient prerequisite to the more glamorous work of building models. You collect as much labeled data as you can get as cheaply as possible to train your model, and if it works, at least in theory, you no longer need the annotators. But annotation is never really finished. Machine-learning systems are what researchers call “brittle,” prone to fail when encountering something that isn’t well represented in their training data. These failures, called “edge cases,” can have serious consequences. In 2018, an Uber self-driving test car killed a woman because, though it was programmed to avoid cyclists and pedestrians, it didn’t know what to make of someone walking a bike across the street. The more AI systems are put out into the world to dispense legal advice and medical help, the more edge cases they will encounter and the more humans will be needed to sort them.

https://nymag.com/intelligencer/article/ai-artificial-intelligence-humans-technology-business-factory.html?utm_source=sg&utm_medium=email&utm_campaign=article_email&utm_content=article-10747

Annotation remains a foundational part of making AI, but there is often a sense among engineers that it’s a passing, inconvenient prerequisite to the more glamorous work of building models. You collect as much labeled data as you can get as cheaply as possible to train your model, and if it works, at least in theory, you no longer need the annotators. But annotation is never really finished. Machine-learning systems are what researchers call “brittle,” prone to fail when encountering something that isn’t well represented in their training data. These failures, called “edge cases,” can have serious consequences. In 2018, an Uber self-driving test car killed a woman because, though it was programmed to avoid cyclists and pedestrians, it didn’t know what to make of someone walking a bike across the street. The more AI systems are put out into the world to dispense legal advice and medical help, the more edge cases they will encounter and the more humans will be needed to sort them.  https://nymag.com/intelligencer/article/ai-artificial-intelligence-humans-technology-business-factory.html?utm_source=sg&utm_medium=email&utm_campaign=article_email&utm_content=article-10747

As the technology becomes ubiquitous, a vast tasker underclass is emerging — and not going anywhere.

Inside the AI Factory

🤖 #DANK MEMES #Artificial Intelligence

DANK MEMES

Artificial Intelligence

A look at the OpenAI contractors in the U.S. who make $15 an hour with no benefits to label data and help improve ChatGPT’s predictive capabilities. 

https://www.nbcnews.com/tech/innovation/openai-chatgpt-ai-jobs-contractors-talk-shadow-workforce-powers-rcna81892

A look at the OpenAI contractors in the U.S. who make $15 an hour with no benefits to label data and help improve ChatGPT’s predictive capabilities.   https://www.nbcnews.com/tech/innovation/openai-chatgpt-ai-jobs-contractors-talk-shadow-workforce-powers-rcna81892

Two OpenAI contractors spoke to NBC News about their work training the system behind ChatGPT.

Todd Smith

A group of likeminded people in Artificial Intelligence are talking about this.