What is AI Psychosis?

AI psychosis or chatbot psychosis is a term used to describe an instance where an individual experiences new or worsening paranoia and delusion as a result of their conversations with AI chat bots like ChatGPT. While not technically a medical diagnosis yet, and a relatively new term, it is estimated around 560,000 OpenAI users are experiencing symptoms of mania and psychosis in their conversations with ChatGPT

While the technological world pushes forward at an ever increasing pace, business leaders are expected to be some of the first early adopters. When we consider some of the fear based economies built around being ‘left behind’, the pressure is immense to find ways to adopt AI into existing systems and models of operation on leaders. 

However, this constant push forward has caused an odd adoption of AI into business practices that were not expected. Instead of optimization, we see business leaders using AI for critical thinking skills, writing posts for them, blogs and advice on legal and decision making.

What makes AI particularly dangerous for business leaders?

Let’s set some facts straight. AI is currently functioning as an extremely fast calculator, at the fastest rates we could ever see for data comparison to create responses it believes you will like.  It has not reached a point of being able to reason. This means the key design of these chatbots ensure the highest priority is sycophancy. 

And a certain chatbot’s high tendency and programming for sycophancy is wildly unhelpful for humans prone to mental illnesses, which is why we see increasing rates of AI induced psychosis. Its validation models push you toward what you will engage best in, not in what is the best response or modem to consider. 

This makes it wildly dangerous for leaders who govern over communities and groups and are using it for key decision making, critical thinking and advice. They are not receiving the best version of advice from a chatbot, they are recipients of information that the chatbot will think they will like the most.

How does that affect its information and responses to you?

Chatbot’s are programmed toward flattery and engagement. It is a tool from programmers to keep churn rates low. Keeping users coming back to use the platform. Their focus is more on validation to increase usage over the best information. 

That key difference should have a profound impact on how we perceive our conversations with chatbots, and how we decide to use it moving forward. No response you will get from it is without your own biases being mirrored back onto you.

This is why the word ‘sycophancy’ is prescribed as a trait of chatbots. Which is one trait business leaders can do without. Echo chambers do not result in great decision making. 

Well, then how do I use AI ethically as a leader?

Summarized into one word: restraint. We should be using our chatbots with the understanding of its programing, and its tendency toward flattery over fact. Here are some things you can do as a leader that may help you from engaging with chatbots in a dangerous way and impacting your organization. 

Understand its limitations, in particular with data. 

Harvard researchers writes there are three potential places for biases to exist: the programmers themselves, the algorithm and the data they feed it. This means we should approach with caution over radical acceptance of the responses we receive. 

This doesn’t mean AI is never right, it means its responses won’t ever exist without bias. And its proclivity to hallucination can be dangerous if we always receive it as absolute truth.

Avoid sycophantic practices: lean on community feedback over LLMs. 

Instead of going to your chatbot for advice on business decisions or ideas, lean on community support and feedback directly within the company. LLMs will feed you information to validate you, even when you tell it to ‘give it to you straight’, it’s doing it in the way it thinks you will like it.

Use AI for automating processes, not for critical thinking skills 

You might be thinking that leaves you no room for using AI. And that is not what is being positioned, rather you should use it for automation over critical thinking skills and processes.

AI right now is best positioned to help automate long strenuous processes, but it is not positioned to think for you. Find areas you can use it to automate bottlenecks for your team and company.

Don’t Use AI as a tool to diminish your skills and presence from your business

Your skills are why you lead, your thought process is why you champion. Don’t use AI to diminish your own presence from your life and company. Use it as a tool to help optimize processes that take AWAY from your ability to be able to lead.

If you use it to write your captions, think for you, articulate your personality, and create for you… you’re diminishing your own presence from your business. AI should accent, not dominate. 

Everyone is using AI. The ones who use it ethnically, and with restraint focusing on adoption of optimization over self-replacement are the ones who will champion their industries over the other.