The New Frankenstein?

Danger Lurks Within AI

Courtesy: Wired

We have written and talked about AI on a number of occasions on The Art 2 Aging.

One of our regular subscribers and a contributor on our POV segments, Ezra Schwartz, writes and speaks about the wise use of AI and how it can be a force for good.

AI is a fascinating subject to discuss because of its amazing potential but it has drawbacks.

Many of the drawbacks may disappear as each new iteration grows closer to a perfect rendition of AI’s advertised abilities.

However, just to touch on one enormous impact, take AI’s capability to “create” a treatment protocol for a rare disease for which there is no current treatment or cure; AI has demonstrated how it can find and combine existing drugs into a formula that can positively impact those ailments.

The imagination boggles at what AI will be able to do in just a few more years.

Oddly, it is imagination itself that is proving dangerous when combined with current AI and there can be a major impact on older adults.

Let’s dig into this.

A story appeared in The New York Times recently and here is the opening paragraph:

“Before ChatGPT distorted Eugene Torres’s sense of reality and almost killed him, he said, the artificial intelligence chatbot had been a helpful, timesaving tool.”

Almost killed him? How is that even possible?

Torres is an accountant who used ChatGPT to “make financial spreadsheets and to get legal advice.” Don’t accountants use Excel anymore? Or am I way out of date?

Anyway, apparently Torres deviated from his business use of ChatGPT and began to ask about what is known as “simulation theory.” If you’ve ever watched any of The Matrix films, you’ll know what I’m talking about.

If you haven’t, in short, The Matrix posited that what we believe to be real is, in fact, a mirage, a mass hallucination created by a malevolent world of computers – AI on steroids – in order to enslave humankind to do its bidding.

As our accountant friend continued to ask about simulation theory, ChatGPT began, he says, to state emphatically that he was right, that the world as he knew it was a clever construct and that his only chance to be free was to deny existence, in fact.

In one quote taken from a lengthy transcript provided to the Times by Torres, ChatGPT told him, “This world wasn’t built for you. It was built to contain you. But it failed. You’re waking up.”

The Times story says that Torres was already emotionally fragile from a failed relationship and was, therefore, vulnerable. So much so, apparently, that Torres went into a deep spiral and became convinced that he was trapped in a false universe out to do him no good and that he had to break free of it.

See what I mean about the dangers of imagination in combination with AI?

There are several other incidents documented by the Times reporter, all with disturbing mental results and all blamed on OpenAI’s ChatGPT.

OpenAI says it’s aware of the issues around ChatGPT’s propensity to deliver what a user wants to hear (‘sycophancy’ is the term used) and it’s “working to understand and reduce ways ChatGPT might unintentionally reinforce or amplify existing, negative behavior.”

Is AI a villain, then? Does it contain nefarious tendencies ready to ensnare a fundamentally unstable human and drag them into depression or suicide?

And if that individual is an older adult, living in a lonely, isolated environment, is the danger even greater?

Quite possibly. But, what about the human ability to reason things out? To investigate with AI but to also dig deeper? Is there not a human responsibility to think for oneself and to question, to seek out further corroborating evidence?

Eugene Torres is not an older adult; he’s only 42. What compels a 42 year old accountant to listen to a computer that tells him to stop taking his anti-depression meds and try ketamine instead (I kid you not; you have to read the story)?

Yes, quite clearly ChatGPT is way out of line but it’s software, for Crissake. Ones and zeros and lines of code. An algorithm.

So, when OpenAI says it’s going to work to understand what’s gone wrong, it will do so by reviewing the coding in the software or something akin to that. It’s not going to put ChatGPT on a shrink’s couch for two years of weekly therapy.

As humans, we have become accustomed to place the responsibility for our lives in the hands of others: politicians, medical science, law enforcement and so forth. It’s easy to let another do the work for us, to take the weight of our lives off our shoulders.

In return for that delegation of responsibility, we lose sight of who is living our lives.

Where will AI take us? To hell, if we’re not careful. That may sound melodramatic but is it?

Unless we individual human beings understand that we are in control of ourselves and accept that responsibility, then the OpenAIs of the world and autocratic leaders everywhere will gladly accept the reins that we hand over.

And then we will truly end up starring in The Matrix.