Beware the AI Mirage: Impressive Speed, Alarming Errors, and the Growing Threat of Fakery

AI is all the rage. But is it all that it's made out to be?

Photo by Igor Omilaev on Unsplash

AI is unreliable

Artificial Intelligence, usually abbreviated to AI, is regularly in the news. I am sure much of the technobabble is pure hype. After all, a year or two ago, many computer programmes, which are still in existence today, are now called AI.

I find that many writers are using AI to create their stories. I really shouldn’t refer to them as their stories as AI can generate stories, fact or fiction, by telling a programme what you want. In a split second, it will churn out the number of words you have requested.

Here’s an example in realtime to show you how quickly a poem can be generated. I asked Grok to write a twenty-line poem about my town, Trowbridge. Click on the video to watch.

At the top of the Google search page, you will now see a summary of your question produced by AI. It looks impressive with the words and answers it churns out so quickly.

Then there are the photos and even videos that AI programmes can generate. They do look lifelike and are getting more so by the day. The systems that can impersonate people, both living and dead, in voice and appearance are impressive. However, these are so dangerous. Just think for a moment of the damage that can be done by such fakery.

Have you tried talking to a chatbot? You know, those pesky pop-up boxes on company websites. Businesses now want you to communicate with these, rather than to be able to telephone, or if you do telephone, forced to talk to a generated voice, that might even have been given a personal name.

The pretence is awful, and I, for one, hate it.

Here’s a conversation I had with one such telephone system.

The system kept guessing what I wanted, and what I needed obviously wasn’t in its repertoire.

After getting nowhere, I asked, “Can I speak to a human?”
… “I’m sorry you’re asking about … (random option)

“Can I speak to a human?”
Same response with a different option.

I asked again another two times, more options.

“Can I speak to a human?”
“I’m sorry, I don’t understand. I will put you through to one of my colleagues.”

How bad is that?

Let me get back to Google. By their own admission, recently, to the BBC, we shouldn’t rely on the AI summary in their search results. If that’s the case, why should the result be there? Here’s the article.

I have found many errors when researching for my stories here at Roland’s Travels & Notebook, as well as on other occasions. Some of the errors are extremely bad, and if these systems had one iota of intelligence, they would not have produced the results shown.

I would also like to demonstrate in the photo generated by Grok how stupid these AI tools are.

A Grok generated photo

I asked for a photo of a UK motorway with traffic. At first glance, it seems good, and the speed that photos are generated is incredible. There is a but. Look closely. Can you see vehicles heading for each other, instead of going in the same direction, and opposite directions between the two carriageways? If Grok had any intelligence, it would know that this photo is ridiculous. There are other issues too with this photo. Can you spot them?

Photo generated by Grok

So, dear reader, be careful. Don’t get sucked in with the hype of AI. Check and double-check your sources of information. Fake news gets even easier to fake. Facts can be mixed up by AI systems and produced in a way that seems so plausible. Check the references, not the summaries.

What are your thoughts about artificial intelligence? Let me know in the comments, and please do share any experiences with it. If you’re not a subscriber, please use the button below. It’s always free to read my stories but if you would like to become a paid subscriber it will be appreciated to support my writing.