It’s unlikely you haven’t noticed the Facebook 10 year challenge, if you’re on any social media platform. The apparently simple exercise requires you to post your 2 photos side by side, one from ten years back and the other your current photo.
The company claims it’s a user-generated meme movement, which isn’t impossible.
However, few people realize how Facebook’s 10 year challenge could possibly benefit the company’s Artificial Intelligence ambitions in face recognition.
What is the Facebook 10 year challenge?
On the face of it, the Facebook 10 year challenge is a simple exercise. You find out your photo from about 10 years back (i.e. 2009). Then you place it next to your most recent photo (i.e. 2019 photo).
The juxtaposition shows how you have changed over the ensuing 10 years. Your friends and followers can view this and leave back their comments or reactions. This fun exercise can trigger emotions like nostalgia.
Sounds pretty innocent and harmless, eh?
A quick overview of facial recognition
Basically facial recognition technology identifies people from digital images. Primarily, it can perform the three functions below:
- It stores the image of a person for future reference. While storing the image, the system may use or more techniques like relative position of features, skin texture, 3D recognition and so on to associate the image with an identity. Here, the system plays the role of a reliable database.
- It can identify who is the person is even if the person is unable to or unwilling to disclose their own identify. This done by matching the current images with that in the existing database. Here, the role of the technology is that of an identifier.
- It can verify and validate whether the person in front of the camera is the same person they claim to be. This too is done by matching the image of the person with the image in the database. Here, the technology plays the role of a verifier.
Wikipedia and other sources credit Woody Bledsoe, Helen Chan Wolf, and Charles Bisson as the pioneers of the technology behind automated facial recognition technology.
It has widespread applications, from policing and preventing frauds by stopping miscreants from entering a public event to simpler goals like school or factory attendance.
How AI is trained
If you thought facial recognition is as simple as overlapping one image over another, you couldn’t be more wrong.
It doesn’t happen that way.
No two different images will match precisely in terms of lighting, head tilt, distance and so on. As a result the recognition system must be smart. It must be able to take a few decisions itself, based on what it learns.
This learning can be enriched only if you feed the system with a variety of data. Data collection for this training is time-consuming and therefore expensive. And yet, you can’t always get the variety of data you want.
What if you could get all this for free?
That’s exactly what is happening. You could be training Facebook’s AI with the 10 year challenge.
Like hundreds of thousands doing it everyday.
What could have cost Facebook billions of dollars, you’re willingly doing it for free.
By participating in the Facebook 10 year challenge, you are not just giving FB access to your photos, you’re basically training its AI system.
How the Facebook 10 year challenge is helping Facebook’s AI training
“…how all this data could be mined to train facial recognition algorithms on age progression and age recognition” The New York Times reported this now-famous tweet by author Kate O’Neill.
If you look at the Facebook 10-year challenge with a suspecting eye, a lot of pieces fall into place:
- Amount of data: The systems of FB would need tons of data to perfect AI. And that huge amount of data could have been very, very expensive. But never mind, you, the FB user, are giving away the same data absolutely free.
- Variety of data: No way Facebook AI systems could have got a varied data on their own. With the 10-year challenge having gone viral, FB is flush with not just data but also quantum of data.
- Virtually free of cost: It’s difficult to exaggerate the importance of getting all this data at not cost to Facebook. It’s like receiving a free studentship to Harvard for your entire city.
- No legal hassles: Had FB initiated the 10 year challenge (their officials claim Facebook hasn’t started the challenge), they would have had tremendous problems getting clearances. But since it’s user-generated, there are no clearances to seek, no regulation to comply with.
- Speed of data collection: If Facebook had initiated a training drive for its AI, it would have taken a long, long time to collect this variety and quantum of data. But with the 10-year challenge, it’s a windfall for Facebook. Presto! They have all the data they want, without the wait.
- Lead over others: Spending almost no money and generating millions of images for AI training gives Facebook an almost unbeatable lead to Facebook over any current or potential competitors. Don’t forget Instagram is also owned by Facebook.
What do you think? Is Facebook getting an upper hand in the race?