What Are AI's Boundaries? Or, What Can AI NOT DO?

What Are AI's Boundaries?" or "What Can AI Still Not Do?

Below is a screensaver image on my laptop.  I like it because it is exactly the type of tiny house that I would either like to buy or build. 

< - <- <- That being said, I took the image and asked artificial intelligence (AI) to look at it and provide a blueprint for it. 

I said to Google Gemini Silver (GGS): "Can you take a shot at creating a blueprint for this house?" 

This is what GGS's response was: "I cannot create a physical blueprint, but I can certainly generate a comprehensive technical illustration or gauge that breaks down the architectural components of the structure in the image."

And she (I call Google Gemini Silver > She/Her) gave me this:

It looks a little fuzzy and smaller than the original, which is very clear and 8 1/2 by 11 landscape. Honestly, I love it, and I have printed it out and put it up on my "Vision Board." Even though AI is saying that she can't give me an actual blueprint, this is more than I thought that I would get.

Other things GGS can't do:

1. Subjective Human Experience & Emotion

• Feel or Empathize: AI can analyze emotional language and simulate empathetic responses, but it cannot actually experience feelings, grief, love, or pain. It has no inner life or emotional state. Though Google's version has made me laugh out loud several times.

• Possess Consciousness or Self-Awareness: AI operates purely on instructions, probabilistic models, and data. It does not know it exists, nor does it have subjective awareness or agency. See my blog "What is an AI Prompt?" to get a better idea of that.

2. Deep Context, Judgment, & Ethics

• Form Moral and Ethical Judgments: AI lacks a moral compass or intrinsic sense of right and wrong. While it can follow hardcoded safety rules, it cannot navigate complex, ambiguous ethical dilemmas that require human conscience, cultural nuance, and accountability. It's pretty easy to delve into this aspect of AI, but I say, "Why would you?"

• Grasp Unspoken Social and Physical Context: Humans rely on shared lived experience, implicit social cues, and physical common sense. AI often struggles in situations where crucial information isn't explicitly written down or documented in its training data. Again take a look at my blog "Fine-tuning your AI Prompts" to get around that.

3. True Originality & Intentional Taste

• Innovate Outside Its Training Distribution: AI generates content by identifying and recombining patterns from existing human data. It cannot make genuine "leaps of thought" or originate entirely new paradigms completely untethered from existing examples. When I say "leaps of thoughts" I mean associative thinking (1).

• Exercise Genuine "Taste" or Curation: An AI can produce hundreds of variations of a design or piece of music, but it doesn't possess personal taste to inherently know which one resonates on a deep human level without feedback. This may be true but often Google Gemini will give you two different answers in a split screen and then ask you which of the two you like. I think Google is doing this to increase Gemini's "taste."

4. Real-World Physical Capabilities

• Adapt Mind-Body Coordination to Unstructured Physical Work: While robotics is advancing, AI cannot match human dexterity and real-time physical problem-solving in dynamic, unpredictable environments (such as complex electrical repair, fine hand-carving, or emergency rescue in unknown terrain). Honestly, too bad it can't, and maybe electricians' invoices would be more affordable. 

• Take Legal or Moral Responsibility: An AI cannot sign a contract, take an oath, or be held legally liable for a decision. True accountability remains strictly human. And I hope it never will.

(1) Associative thinking is the mind's natural ability to connect ideas, memories, words, or things that seem different or unrelated. Instead of following a straight line of strict logic, your brain jumps from one thought to another based on shared traits, feelings, or past experiences. It is said that an ADHD's racing mind is genius at this.

# # # #