By clicking “Accept”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.
November 28, 2023

Newsletter - How we work with Gen AI at isahit

November 28, 2023

How we work with Gen AI at isahit

This year has marked a significant breakthrough for the emergence of Generative AI technologies. Initially, we were concerned that generative AI could be a threat to our core business of data annotation.

However, as we encountered numerous diverse use cases and witnessed the continuous evolution of datasets, along with the wide range of very specific needs, the past months have confirmed what we are certain about for years: Human expertise plays a crucial role in the next era of customer-ready AI and the development of LLMs.

Our labeling approach combines AI and human intellect, balancing technology and human feedbacks. It’s time for us to show you how we deal with Generative AI and LLMs at isahit!

At isahit, we use various tools to label your data.
In addition to these tools, we also incorporate Generative AI tools into certain client projects to maximize the effectiveness of our human labeling efforts.
Some examples of these tools include ChatGPT, Microsoft Azure OpenAI, and Llama2. These tools are used to enhance the annotations provided by our labellers.

Or vice-versa! 🔃
Sometimes, our Human taskforce needs to control the generated content, particularly content produced by our Language Models (LLMs) and GenAI. Here is an overview of how we integrate LLMs and GenAI at isahit ?

  • We collect and pre-train your datasets:
    ​​Starting from scratch, including the research of sources, evaluation, and labeling of raw data.
  • We label your data:
    To build strong LLMs, you need to get queries and prompts right tagged for better dialogues between humans and machines: classification, tagging, suggestions…​​​​​
  • ​We correct generated data:
    We use RLHF and HITL to evaluate large language models, ensuring accurate output. This expertise improves the accuracy of your AI and machine learning models through verification, evaluation, and correction of your prompts and generated content.

📣 Testimonial from a Human in the Loop: Laetitia

Laetitia testimonial

​​​​​​
Discussing the importance of Human-in-the-Loop in the data labeling process, it is essential for us to give a voice to our labellers. This month, we encourage you to view Laetitia's testimonial from Madagascar.

You might also like
this new related posts

Want to scale up your data labeling projects
and do it ethically? 

We have a wide range of solutions and tools that will help you train your algorithms. Click below to learn more!