Human or AI?

Posted on by Tim

“Even a manically depressed robot is better to talk to than nobody.“

~Douglas Adams

Following up on last week’s blog “Ironic Conundrums,” I again turned to OpenAI’s Chat-GPT-4 system for assistance.

My reasoning for doing this is simple.

Artificial intelligence as a business, societal, governmental, educational, and personal tool isn’t going to go away.

It is only going to continue evolving faster than we can implement regulations and controls governing its usage parameters. It is absolutely an integral part of the future of our #Culturalinertia, for better or worse. Humans have a long history of becoming dependent upon technology to the point of unlearning how to do the things that technology was built to assist them with.

With this in mind, we need to learn how to ensure we are steering toward better as much as we possibly can.

To help us do that, I asked GPT to “Explain what steps an average person should be able to take to determine if something they are reading was written by an artificial intelligence text generator.”

Following is its response. Don’t miss the irony in Step 10, which advises us to ask AI if we are being tricked by AI:

Determining if a text was written by an artificial intelligence (AI) text generator can be challenging, as AI-generated text has become increasingly sophisticated. However, there are certain steps an average person can take to identify AI-generated content:

  1. Look for inconsistencies: AI-generated text may have inconsistencies in tone, subject matter, or point of view. If the content seems disjointed or switches between topics without clear transitions, it might be AI-generated.
  2. Analyze grammar and syntax: AI-generated text may have unusual grammar or syntax. While it has become less common in advanced AI models, it can still serve as an indicator of AI-generated text.
  3. Evaluate coherence and logic: Check if the content flows logically and is coherent. AI-generated text might sometimes produce sentences that sound plausible but lack a clear connection to the overall topic.
  4. Fact-check: AI-generated content might contain inaccuracies or outdated information. Fact-checking the text against reliable sources can help identify if the content was generated by an AI.
  5. Repetition: AI-generated text might repeat phrases, sentences, or ideas throughout the content, which could be a sign that it was generated by an AI.
  6. Unusual word choice: AI-generated text might use unusual or rare words, sometimes inappropriately or out of context. This can be an indication that the text was generated by an AI.
  7. Check for overuse of common phrases: AI-generated text may rely on common phrases or idioms more frequently than a human author would. If you notice an overuse of clichés, it might be AI-generated.
  8. Emotional connection: AI-generated text might lack a personal touch or an emotional connection. If the content seems impersonal, it could be a sign that it was generated by an AI.
  9. Source credibility: Consider the source of the content. If it’s from a reputable website, organization, or author, it’s more likely to be human-written. Unfamiliar sources or those with a history of sharing AI-generated content should be approached with skepticism.
  10. Use AI detection tools: Several online tools have been developed to help detect AI-generated text, such as OpenAI’s GPT-3. These tools analyze text for patterns associated with AI-generated content and can provide a probability score that estimates the likelihood of the text being AI-generated.

Keep in mind that these steps are not foolproof, and some AI-generated content can be very convincing. It’s important to remain vigilant and use critical thinking skills when evaluating the credibility and authenticity of any content you encounter.