LinkedIn’s taking a novel approach to training its AI tools, by enabling LinkedIn members to participate in the data ingestion process, a task that’s normally been left to armies of contractors who sort through and label massive data sets.
But on LinkedIn, you can get paid to do it, using your industry expertise to help improve LinkedIn’s, and likely parent company Microsoft’s, AI tools.
As explained by LinkedIn:
“LinkedIn is offering members an opportunity to earn flexible, skill-based income using their expertise to help companies develop high-quality, human-labeled data for AI training. This initiative reflects our commitment to creating economic opportunities for every member of the global workforce.”
As noted, many AI projects use cheap labor to label datasets, in order to help guide AI tools in responding to queries. But LinkedIn’s looking for more specific, specialized, industry-specific insight, via people who’ve worked in specific roles and areas.
“While any LinkedIn member can express interest in becoming an AI trainer, we must confirm that you have the expertise to complete annotations as indicated by your profile.”
For this, LinkedIn says that it will conduct interviews with prospective AI labellers, in order to determine if their skills and experience are a fit.
“For projects related to your specific expertise (e.g., law, medicine, finance), LinkedIn will use your profile data, such as education, licenses, and work experience, as well as an AI-powered informational conversation where you can share your experience in more detail. We will use this information to match subject-matter experts to the right tasks.”
So it will use an AI system to interview about your background, in order to determine if you are then suited to help train its AI project.
Which feels a little dystopian, but…
“This feature uses AI to ask you questions about your professional background and respond to your answers, simulating a realistic conversation. Your conversation data and related AI insights will be used to supplement the information already on your LinkedIn profile so that we can better match you to relevant AI training projects based on your experience and expertise and assess your fit for specific projects. We will also use this information to suggest updates to your LinkedIn profile. We will not use this information for other purposes without your permission.”
I guess that’s a bonus, that the process will also tell you why your LinkedIn profile sucks, and how you can improve it.
The process of annotating data like this enables AI systems to better understand what each element means in specific context. So in this process, LinkedIn’s looking to ensure that AI tools can understand how different industries and professions refer to different products, tools, and other elements, which will then enable LinkedIn to improve its data matching tools, and maximize its own AI recommendations and guidance.
Though as LinkedIn notes, this also relates to data training for other companies as well, with these experts being used to provide specialized insight to assist in AI training.
Which means that by participating in this project, you could also be training AI systems that might take your job, or others in your industry.
If you’re okay with that, then you can get paid for your contributions, via LinkedIn.
It’s an interesting approach to model training, using LinkedIn’s reach to a broad range of professional industries to recruit people for the task.
LinkedIn says that members can opt in to receive notifications about upcoming annotation projects, and once vetted, they’ll automatically be matched to opportunities that fit your skills and availability.
You can read more about LinkedIn’s AI annotation process here.
