AI UPDATE: Human Labor Makes AI Possible, 18 Use Cases For Midjourney, And Building A Financial Analyst

The Hidden Human Labor Powering AI: Unmasking the Unseen Annotators

In the shiny, ultramodern tech world, where AI chatbots charm us with their witty responses and autonomous vehicles promise a future of leisurely commutes, a largely unseen workforce toils away, far from the spotlight. They're the data annotators, the unsung heroes who make the magic of AI possible. And their story is a stark contrast to the glossy narrative often spun about AI's potential.

Meet Joe, a 30-year-old from Nairobi, who found himself in the labyrinthine world of data annotation after college. What started as a tedious job of tagging raw data to train AI systems turned into a boot camp management role, instructing 50 new recruits every two weeks. The work was monotonous and grueling, often leading to a significant dropout rate. Despite the tedious nature of the work, it was a job in a place where jobs were scarce, and Joe was able to turn out hundreds of graduates.

This workforce operates under a veil of secrecy, with many annotators unaware of the exact purpose of their work or even the identity of their employers. In Joe's case, the company employing them was Remotasks, a subsidiary of Scale AI, a multibillion-dollar Silicon Valley data vendor. The workers rarely knew the exact nature of their work, which included a wide range of tasks from labeling objects for self-driving cars to categorizing distorted dialogue snippets. Their work has an undeniable purpose, but that purpose is often hidden from them, a fact that adds another layer of ambiguity to their job.

The foundation of the AI boom lies in this ocean of human-labeled data, which was realized by Fei-Fei Li in 2007. By harnessing the power of crowd-sourced labor via Amazon's Mechanical Turk, she was able to assemble a massive annotated dataset that kickstarted a decade of progress in AI. Despite this, the annotation process is often dismissed as a temporary inconvenience rather than a critical part of AI development.

The annotation tasks range from classifying the emotional content of TikTok videos to identifying corn for autonomous tractors. These annotators are the invisible workforce that keeps the AI running, a testament to the continuing need for human intervention in the world of AI. As Sonam Jindal, the program and research lead of the nonprofit Partnership on AI, stated, "Human intelligence is the basis of artificial intelligence"

However, these annotators are not merely cogs in the machine; they are individuals whose work conditions warrant serious consideration. Appen, a publicly traded Australian company, lists 118,200 people performing AI-related tasks, but they are considered independent contractors, not employees, raising questions about their working conditions and pay. The work they do is incredibly detailed and granular, yet their rights and protections remain in the shadows.

There's an urgent need for transparency and accountability in this sector, not only for the sake of the annotators but also for the future of AI itself. The AI community must ensure that the very human labor that powers its innovations is treated with the respect and dignity it deserves.

The next time you marvel at the latest AI innovation, remember the hidden human labor behind it. It's high time we gave credit where credit is due. After all, as AI continues to shape our world, we need to ensure that the process doesn't lose sight of the human element that makes it all possible.

Scale is one of the leaders in annotation, founded in 2016 by then 19-year-old, Alexander Wang. In 2021 it was valued at $7.3 Billion but has fallen some since.

You can take a deeper look into the labor behind AI in this article from Verge.

18 Creative Use Cases with Midjourney AI Art

In the world of art and creativity, the emergence of AI has brought forth exciting possibilities. Midjourney AI Art is at the forefront of this revolution, empowering artists and enthusiasts to explore new realms of imagination.

In this article, Christie C. dives into the 18 Creative Use Cases of Midjourney AI Art, showcasing its versatility and the remarkable artwork it can produce.

From surreal landscapes to professional mockups and photorealistic images, let’s explore the extraordinary ways Midjourney AI Art is reshaping the artistic landscape.

Check out her article on Medium…

PROGRAMMING: Building an AI Financial Analyst Leveraging OpenAI’s New Function Calling

Last week, OpenAI made a bunch of announcements including function calls, an extended context window for Chat-GPT 3.5 turbo, and a cost reduction for the 3.5 model as well.

I am starting to see more projects showing up leveraging the new function calls functionality including this example of creating an AI Financial Analyst.

His first example was repetitive and not very useful, but by leveraging the functions, the second round provided a much better analysis

Generating Random Image Prompts For Midjourney

Lately, I have been generating travel photography shots for my Facebook Profile header image using travel spots I already knew. I thought it would be a fun exercise for ChatGPT to come up with them by choosing one at random and filling in the prompt.

This prompt works best with Midjourney 5.1 and using the —style raw option. Both of these can be set in your Midjourney settings.

PROMPT:
Generate a random image prompt for Midjourney AI for a random natural tourist attraction using the following output template:

/imagine Breathtaking {random natural tourist attraction}, shot with Nikon D850 and Nikon AF-S NIKKOR 14-24mm f/2.8G ED lens, national geographic --ar 16:9

Images Of The Day

Prompts for all of these images were generated using the ChatGPT prompt above, with ChatGPT picking the natural tourist attraction. I plan on creating an automation to generate images for Instagram profile using a similar process.