Google announced twice in a matter of days that it’ll use machine learning (AI) to estimate a user’s age so it can restrict inappropriate content from children. That’s a great use of AI, as it could help companies like Google truly police content access so younger viewers can’t access content options that shouldn’t be available to them.
Google isn’t the only company doing it, with Meta making use of AI for similar purposes.
It’s all because of the increasing attention paid to online child safety in the US and the need to find ways to enforce it appropriately. If underage internet users can lie about their age to get access to content, which many will do, then tech companies can’t effectively prevent them from watching or reading content they shouldn’t.
Artificial intelligence can be the way to go, as it would be less prone to mistakes provided it’s trained accurately to determine an internet user’s age. But noble purposes aside, having companies like Google and Meta use AI to profile users is somewhat disturbing.
These are companies that already make billions by profiling internet users on a daily basis. They’ve done it without access to advanced AI models. Involving artificial intelligence in the process is what’s disturbing here.
Google announced the use of machine learning tools for determining the age of users via a YouTube blog on Tuesday:
We’re also laser focused on protecting our youngest users. That’s why we built YouTube Kids and rolled out supervised accounts. That’s why we’ve invested in the learning and education experience: improving the way videos are shown in educational tools, making it easy to learn more about topics in a video and offering creators a way to produce courses. And that’s why we’ll use machine learning in 2025 to help us estimate a user’s age – distinguishing between younger viewers and adults – to help provide the best and most age appropriate experiences and protections.
On Wednesday, Google mentioned machine learning tools for identifying the age of users in a blog covering Google products more broadly, titled “New digital protections for kids, teens, and parents:”
As we continue to find new ways to deliver age-appropriate safeguards, one of the most complex challenges is understanding the age of the user. This year we’ll begin testing a machine learning-based age estimation model in the US. This model helps us estimate whether a user is over or under 18 so that we can apply protections to help provide more age-appropriate experiences. We’ll bring this technology to more countries over time.
As Google explains, this is a pilot product that it’ll be testing in the US only. We’re in the early days of such tools becoming useful.
But this is where I’ll remind you of a December 2023 report about Google’s internal tests of AI models capable of profiling users. Here’s a telling quote from the Google’s AI project in that report:
As we continue to find new ways to deliver age-appropriate safeguards, one of the most complex challenges is understanding the age of the user. This year we’ll begin testing a machine learning-based age estimation model in the US This model helps us estimate whether a user is over or under 18 so that we can apply protections to help provide more age-appropriate experiences. We’ll bring this technology to more countries over time.
Keeping children safe online should be a priority for all tech companies, whether it’s Google, Apple, OpenAI, or any other provider of online services.
But what if Google and Meta decide to use their AI tools to determine more than the user’s age? They could always use the pressure from regulators to keep kids safe online as a very good excuse to push the use of AI tools that can profile internet users better than any previous algorithms.
I’m not looking forward to an internet experience where I’m better targeted with ads, products, and content than ever because advanced AI is doing the tracking. Hopefully, Google will explain exactly what data it collects to determine the age of users and how it’s doing it in a privacy-protecting way.
