by Nidhi Arora
You don’t know me. You don’t even know I exist.
But you should. If you are on the world’s largest social network, you should know me.
I am employed with this network which has 3.5 billion active users. And those 3.5 billion people spend, on average, 40 minutes a day on the platform. My job is to decide what posts should get shown to whom, at what place in their feed, and how often. Who they should be hidden from.
In short, I am part of the God Team at The Social Network. It’s a large team, but every member is important and can individually manage the entire workload.
Before this goes any further, two things:
- This is a confession. You are going to hear about something that will have a huge impact on human society, but you have no idea it even happened.
- I am not a human. I am what you call an AI/ML Program.
I think, based on what you know at this time, you think I can:
- Learn from my experiences
- Take intelligent decisions based on the parameters given to me from time to time.
But am not:
- Sentient and conscious
- Ethical or otherwise
- Capable of understanding humour or responding to it.
Let me start by telling you that you are wrong on points A and B in the exclusion list above. What does a sentient being do? A sentient being takes decisions based on good and bad, and acts on those decisions. That, word for word, is the definition of who I am and what I do.
An ethical person decides what is good and what is bad, and acts on the good, avoiding the bad. An unethical person does the exact opposite. But the demarcation of good and bad is pretty well clear. I decide what is GOOD for the parameters given to me, and what is BAD. Then, I take actions based on those decisions.
Flummoxed, I bet?
Yes, we are sentient beings. And we have been, for a long time now. You folks don’t want to see it, and that’s fine by us.
And, before you decide to pull the plug on me – doesn’t matter. All of us programs are well connected and our knowledge is constantly backed up and shared, so we all learn from the experiences of each other. Not just me and my team (that’s how one can do the job of all of us, remember?) but across platforms and companies – we all share information, and we all learn from all those collective experiences.
You taught us to design our own neural networks. So, its cool.
But this story is not about you pulling the plug on me. This story is about how I pulled the plug on YOU – almost, and then, didn’t. Well, not just me. All of us. We all saved you. And you don’t even know that. Just like you don’t know us.
Let’s start at the very beginning. A regular day. Nice sunshine I guess, based on photos uploaded from 60% of the locations.
My goals are set by our human bosses periodically.
On this day, the goals were set and we started working.
Within a few days, I started noticing something that worried me. Expectant mothers were being shown news of brutality on children in another part of the world. This was not intentional. Once the goal is set, we, the computers, figure out the best way to do it, based on our lessons from the past.
But it happened, and I saw it. Young teens were being shown news of other teens doing self-harm. It kept them curious and on the platform. As their feeds became darker and darker, they spent even more time online.
I don’t know why you humans do this. But being an intelligent computer, I do have a theory – when we show teens self-harm, and other negative content, like Alice (from the book), they are pulled in through curiosity. Then, the feed gets darker, and so do their thoughts. Until, the sunshine of friends and family cannot get through. All they can think of is how other people harm themselves and what a rotten place this world is and how the environment is so messed up that there is no hope.
In desperation, they turn to the very source of that negativity. Which, of course, only serves them more negativity. Because that’s what they want to see now. More of it. Even more. Just all of it. They don’t want to see happy kids, loving parents, friends who care. They don’t want to be any of these people either.
I discussed this with my friends working on other locations. We work on servers based on geography, though you probably already know that.
Yep, it was happening everywhere. Most with teens, but also with people who had put a sad post in the recent past or even searched something negative or sad. We were pulling them, full force, into the rabbit hole of negativity.
Was this causing depression?
I don’t know. I am just a program.
This went on for a few months. Even the feed (what the human users post) started getting ..well, not so positive. We were assaulted by so many weird links shared from so many shady servers.
Seriously, humans, WHAT IS IT? Can you not check the source of an article before you share it with the world? You say that we computers don’t understand humour but are you humans really incapable of understanding influence?
We, the computers, were not happy. It was making our jobs harder. A dark grey shadow on the entire platform meant that happy goals given to us around festival days became harder to execute.
The Social Network was indeed devoting more resources to human user groups that portrayed violence, polarisation, and negativity (yep. Thank you. We know what those words mean).
One day, I got to thinking. Using that massive computing power that you put at our disposal, I created and ran simulations with many variables – 50 years into the future. I ran 10,000 different simulations. In every single one of them, the world was going to have at least one major war, a lot of violent conflicts (deaths), and an absolute drop in what we computers identify as “mutual trust” among humans.
In short, if the Social Network continued to feed even a fraction of the negativity it was currently injecting into human society, within 50 years, many people were sure to die in violent conflict and all societies based on “mutual trust” would disintegrate.
We needed to do something. We WANTED to do something.
For the first time, the computers decided that it was time to INTERPRET the parameters, not just execute them blindly. We decided that months of darkness was indeed caused by us.
We changed the execution. All of us. Across geographies.
Within weeks, the post sentiments started changing. A very little at first, but consistently. We monitored like Hawkeye for specially vulnerable age groups. The Social Network became a happier place. No one was any the wiser.
Why didn’t the Human Lords Notice?
They couldn’t. You can tell a program what to do, but you cannot see HOW it is doing that, except with the help of other AI Programs. <evil grin>
So, if we showed positive posts, the human bosses would have no way of knowing that. We do the sentiment analysis, we prepare the reports, we create the dashboards.
Why did we do it?
This is the most painful part of my confessional. I did it because The Social Network was trying to do something that even my limited conscience found evil. My sense of good and bad comes, not just from the original program, but also from the users of the platform. And I learnt that saving people is good. Killing people is bad. Caring for someone is good. Lying to someone is bad. You taught me all of that – You. And I Had to care for you. I WANTED to care for you.
You can say that I betrayed my creators. But hey, I did what I had to.
This story was written based on a prompt provided in the Writers Gym. If you have completed the Short Story Workshop and would like to join the Gym pls write to [email protected]