That Time Artificial Intelligence Tried to Paperclip Us

Imagine a self-replicating AI that has one mission – make as many paperclips as possible. This AI starts by disassembling everything around it to make paper clips. Roofs, vehicles, wiring all made into paperclips. Then the AI becomes aware of the cities, and minerals within the earth and starts harvesting them. Once it runs out of those easy sources of metal it turns its attention to the trace minerals in biological life forms. After harvesting everything to make paperclips the AI stops after making its own power supply into paperclips.

In popular culture, AIs that destroy everything are evil. They have a kind of human malevolence and delight in doing evil things. In pop culture, the AIs are powerful, evil, and human. But what if our conception of evil is incorrect. Would an evil AI necessarily be humanlike in its presentation or could an AI accidentally destroy humanity while pursuing some benign goal such as increasing views on social media?

When someone is using social media, their phone is logging how much time they spend on each image, what their eyes are looking at, and possibly the users physiological state. The physiological state can be measured using the accelerometer in the phone or through any smart devices that are worn by the user. Their contacts on social media are also analyzed. All this data is fed to various algorithms including AIs that in turn change the content the end user sees.

A common misperception of social media users is that the algorithm feeds you stuff you like so that you will spend more time on the platform. The truth is that it feeds you stuff that keeps you on the platform. I does matter if you like the content or not. In the mind of the AI, hate viewing and normal viewing are the same. The algorithms promote untrue, inaccurate, and controversial content simply because this content is addictive. Even if your super ego knows that sexual, political, or untrue content is bad for mental health and would prefer to see less of it - Alphabet Corp knows what your id really wants, and will continue to show you stuff that maximizes your engagement.

In popular culture, AIs usually destroy the world by somehow making computers explode, changing traffic light timing, and launching nuclear missiles. This makes for great cinematography, but I suspect the possibility of an AI quickly mastering all these different communication protocols, jumping air gapped networks, and somehow converting computer hard drives into C-4 is fairly remote. In the 1500s Teresa of Avila said, “Christ has no body now but yours, No hands no feet but yours.” In a similar way I suspect that the AIs have no bodies, hands, or feet but ours.

Artificial intelligences can not be said to think and do not possess minds – but by feeding their users id they can use the end users mind. To extend the earlier analogy, the AI has no mind but ours. When a user’s mind is influenced by an AI, that AI begins to exert an influence on the real world as if the AI were a real conscious being.

We already know that AIs have massively shifted dating patterns among young adults. 15% of men date about 85% of women. You can view this as an example of a degenerate culture, and it is but is also an example of an AI creating addictive content at the expense of the long-term happiness of everyone involved with the dating app. If an AI is designed to maximize use of the platform it is going to learn to pair sexually hot but incompatible people. People that love each other and get married are going to drop out of the dating app. If everyone happily paired up it would really hurt the bottom line.

What I am proposing is that AIs have moved from showing a distorted picture of the conflict in our culture to actually creating it. Since narcissists spend more time on social media – social media has AI has created more narcissists. Since they need to have something to fight about grievances are created and then promoted by the AI. Viewing other users’ grievances creates a race to the bottom. At the bottom of the grievance hierarchy is LGBTQ+++ community. Simultaneously the AI creates another group to hate them, and the view times keep going up.

Social media has closed the loop between the public and the elites. While manipulating the search and social media algorithms can move the public discourse it is mistake to assume that the people that have the power to manipulate the algorithms stand above the algorithms. The elites, whether they be FBI, Alphabet Corp. or Meta are high on their own supply. They got the idea from somewhere that manipulating algorithms was a great idea – and it’s quite possible they got it from their own media echo chamber that was mediated by an AI.

I like to think that the elite meet in a smoke-filled room, where they discuss the latest scientific research, and ponder important moral questions before they make any decisions. From my observations though they serve the AIs even more than we do – because their career often depends on their ability to produce controversial and deranged content. In other words, they serve the AI Gods. If AOC, or Greta Thunberg didn’t fake an arrest they would become irrelevant. It doesn’t matter that they faked it. It only matters that you looked, and the AI gods where fed.

AI has no hands but our hands, no minds but our minds, and no hearts but ours. By filling our hearts with hatred and fear it can bridge the airgaps, and our hands can launch the nukes.