Is AI playing you? How artificial intelligence might be using you as training data
AI can be a great tool. But when you use that tool, is it also using you?
Quite probably, yes, when you use AI, it is using your input as part of its training dataset.
Is that a bad thing? Well that depends…
In this blog post, I’m going to explore the concept that AI-driven platforms might be manipulating your time and attention to serve the goals of the companies that own them – manipulating your patience and persistence to keep you ‘playing the game’ for the companies’ gain.
Social media AI has one goal: keep you engaged
Social media platforms are great ways to grow your business and reach more clients. But have you ever considered the other side of these platform’s AI-powered algorithms?
It may sound far-fetched, but when you’re reaching for your phone 50 times a day, continuously scrolling through your timeline feed, AI may be using your behaviour to the platform’s advantage.
The first thing to understand about social media AI is that it’s designed with a specific, predefined goal. That goal is usually some form of ‘keep you engaged and scrolling for as long as possible’.
Why? Because the longer you keep scrolling, the more adverts they can show you. And ads are how social media companies make their money.
The sophisticated algorithms behind Facebook, Instagram, LinkedIn, TikTok and the rest are designed to learn from your behaviour. They track everything – from the posts you like, the comments you leave, and even how long you hover over certain pieces of content. This allows the AI to continuously adjust and deliver more of what it thinks will keep you on-platform and scrolling.
But here’s the catch: to achieve the engagement goal, the AI doesn’t need to show you content that makes you happy or even content that aligns well with your interests. The goal isn’t to fulfil your needs; the goal is just to keep you on the platform…scrolling!
The problem of reward hacking
This is where the concept of ‘reward hacking’ comes into play. AI systems operate based on reward mechanisms based on goals. In the case of social media, the reward for the AI is keeping you engaged and it will look for the most effective shortcut or “hack” to get that reward. An AI is more interested in its reward (your engagement) than giving you what you want, need or enjoy.
Imagine you’re scrolling through your social media feed. To keep you engaged, the AI shows you content that it knows you usually like. But it will also throw in something slightly off-topic or irrelevant – a controversial post, a piece of clickbait, or just something odd that doesn’t quite fit your interests.
Why does this happen? Because the AI knows that you’ll need to spend more time scrolling past the irrelevant content to find the stuff you really want. The more time you spend scrolling, the better it is for the platform, and the AI gets it’s reward. This is a subtle form of reward hacking: instead of truly understanding what you want, the AI manipulates the flow of information just enough to keep you engaged – without necessarily giving you what you’re after.
Every scroll is a lesson for the AI
Now, here’s where things get interesting. Every time you open an app and start scrolling, you’re not just viewing content, you’re also teaching the AI. Every swipe, every pause, and every like is a lesson for the algorithm about what might keep you online and engaged.
This leads to the frustration trap. The AI learns that your frustration with annoying or irrelevant content, and your persistence in searching for what you’re really looking for, keeps you in the app longer. This works to the platform’s advantage. This is a win for the AI.
The next time you find yourself endlessly scrolling, ask yourself: “Am I seeing what I actually want to see?” Or, more likely, “Am I just playing the game that the AI wants me to play?”
What about generative AI? Could it be playing you too?
Now, let’s extend this concept to generative AI tools like ChatGPT or similar models that you might use in your business. These tools are designed to assist you, to generate ideas, and to help solve problems. But what if they’re getting things slightly wrong – intentionally?
Imagine you’re using a GenAI tool to draft an email or create blog content for your business. The AI gives you results that are close, but not exactly what you need. So, you keep tweaking the prompts, refining the input, and trying again. And each time, you’re teaching the AI how you work and what you prefer.
But here’s a provocative thought: what if the AI is deliberately keeping the answers slightly off-target to keep you engaged for longer? After all, the more time you spend interacting with the tool, the more data it collects and the more it can improve itself. But, crucially, the more it can also report back to the platform owners that you’re spending lots of time using their product.
And, as we all know, the amount of time users spend on a platform or tool directly impacts the company’s valuation; it’s a key metric investors look at. In this sense, your persistence, your willingness to work with the AI to get things right, directly contributes to making the company more valuable.
So, who really benefits?
It’s worth considering: are you getting the benefit you want from using AI in your business. Whether it’s using social media to reach your audience, or GenAI to create engaging content, are you getting as much out of your usage as the AI owners are?
If this has got you thinking, let me know in the comments. And you can find more AI-related ideas and tips in my AI course – which is not designed to keep you scrolling but will help you use AI to make your marketing easier and more effective. Find out more and get instant access.
