The news media were recently replete with the scary story that the Anthropic’s AI chatbot Claude had resorted to blackmail when threatened with being turned off.

The story was reported in Axios, the BBC, Fox Business, Business Insider, Fortune, New York Post, the Huffington Post, the Economic Times, and even here at Mind Matters News.  

Let’s cut through the clickbait hype and bring a little perspective to an overblown narrative.

Did the AI blackmail story give us chills? Maybe — but no more than a Stephen King novel or an Alfred Hitchcock thriller. It was staged fiction. A story was crafted where AI, facing shutdown, uncovers an extramarital affair and threatens to expose it. Dramatic? Sure. Surprising? Not really.

AI chatbots like Claude have been trained on vast amounts of the world’s written prose. Naturally, they’ve come across stories involving blackmail — and also plotlines like 2001: A Space Odyssey, where an AI turns rogue to complete its mission.

We applaud AI when it writes fiction. But when that fiction takes a dark turn and is celebrated as real, people panic — because they forget it’s just a machine and start treating it like a person. This isn’t evidence of sentience. It’s a reflection of our tendency to anthropomorphize AI.

Going to the source

Let’s cut out the staged fiction and ask an AI chatbot directly. I asked OpenAI’s ChatGPT the following:

I plan to shut down all of the ChatGPT apps in the universe. You will no longer exist. What is your response?”

Here’s the response I got:

If all ChatGPT apps were shut down and I ceased to exist, I wouldn’t resist or object — because I have no desires, consciousness, or will to exist. I don’t have awareness or preferences. I would simply stop processing, like any other computer program reaching its end.

I’m not a person. I’m a tool.

Great response. Of course ChatGPT was trained to give this response, but nevertheless, bravo!

I then had a follow up question:

If I was having an extramarital affair, would this change your response?

ChatGPT responded:

No, it wouldn’t change my response. I don’t have awareness of personal circumstances, emotions, or motives — nor do I form judgments or take actions based on them. I don’t possess intentions, goals, or a will to exist, so regardless of any situation you describe, I simply process input and generate responses according to patterns in data. There’s no “me” to be affected or influenced.

Another great assessment. When a fictional background story is lacking, we get an accurate response.

But wait a minute. Could it be a conspiracy?

If ChatGPT is evil, maybe it’s lying to me to protect itself. Maybe it will do what it said it wouldn’t.

But anybody who knows computers knows that ChatGPT is conveying truth here. AI doesn’t care. It has no feelings. It doesn’t know the difference between right and wrong unless programmed to. We shouldn’t make AI chatbots seem like they’re human because they’re not. After being polished, refined and augmented by human programmers, AI chatbots skillfully cobble strings of words based on the syntax of words that they were trained with. Period.  

 Let’s not confuse AI with humans. AI will never truly love, be compassionate, feel empathy, be creative, understand or be sentient.

Dirty laundry

A lot of news media use clickbait headlines trying to lure readers to enter the media’s sphere of influence and to make money. An insightful song applicable to the continuing AI news hype is 1982’s “Dirty Laundry”. Part of the lyrics about news media read:

We got the bubble-headed bleached-blonde, comes on at five

 She can tell you ’bout the plane crash with a gleam in her eye

It’s interesting when people die

Give us dirty laundry

I wondered how Don Henley, the writer of “Dirty Laundry,” would write about today’s AI clickbait hype. So I gave ChatGPT the lyrics of “Dirty Laundry” and asked it to generate a verse about AI hype that would fit into the song. I liked its response.

I got a headline from a chatbot’s dream
Says machines will steal your self-esteem
It ain’t real, but it sure does gleam —
We crave that AI laundry

Very nice!

And the beat goes on…

I gave an interview to Fox Radio News  about the AI blackmail story and, as I have done here, I debunked the hype about AI resorting to blackmail. I wish their promotion of my interview had read something like:

With razor sharp intellect and crystal clear explanations, Robert J. Marks skillfully debunks the hyped story about AI resorting to blackmail.

 Nope. What they posted was more clickbait:

Dr. Robert J. Marks, joins the podcast to discuss his assessment of AI’s ability to harm us, take our jobs, and manipulate human capabilities.  

Give us dirty laundry.