AI and The Cocoanuts

June 4, 2026 MountainFinchPost

May 31, 2026

As an engineer, I have worked with computational programs and codes for modeling systems and controlling processes. If you weren’t familiar with such programs, you might watch an operating process and assume it is thinking for itself. But this is not true. A program provides the logic, the steps, the boundaries, the formulas and receives the information, from instruments and sensors, from which a decision is made and the process operates as intended. Such programs can even be designed to go outside boundaries which appears to be an even higher level of thinking. I’ve been around a long time and this stuff was around well before me. I call these logic algorithms (focused on process control).

I’ve also worked with supercomputers. In fact, they’ve gotten smaller over the years, to the point where I had respectable capability in a tower on my desk. The purpose of supercomputers was to do increasingly complex calculations quickly. The pharmaceutical industry, among others, has been a heavy user as they explore potential drug molecules for treating disease. The use of supercomputers has been key to the rapid discovery of new treatments and cures.

Anything beyond logic algorithms and supercomputers needs to have certain attributes, significantly better by any performance metric, to seem worth pursuing:

much faster (must be accurate)
smaller footprint
less energy and materials consumption
lower cost per computation (that eventually produces a requested result)
finally, able to mimic a human – think or reason logically, inductively and deductively

Now the last one has been a goal for decades. Thus, the term artificial intelligence or AI. The reason this is desired is because the human brain can solve problems outside of the code (experience and education) it has been provided. This is how invention often occurs. However, the first four are very acceptable as advancements important in problem-solving, decision-making and invention and don’t necessarily require AI. Just doing the first four can accelerate the creation of many new and beneficial methods, systems, materials, devices, processes and technologies.

AI has been pushed aggressively now for several years. Most of the general public never heard of it 5 years ago. Why the rush? Is AI so much better than these other methods that we really need to go after it and for what purpose? There’s only one reason I can think of which might make AI far more desirable than just improving current programming methods and supercomputers:

Is AI able to very rapidly solve a problem or make a good decision

all by itself by just asking it (how about within minutes, hours or days)!?

Let’s try an example. Suppose I ask an AI agent to find a new type of battery that can power a modern cell phone for 1 month without requiring a re-charge. Typically, re-charging is required within a few days of use. This is a problem that might be worth solving. Now there’s a few caveats here:

the battery must be a practical size for a phone
the battery must be manufacturable (it can’t be made of unobtanium)
the cell phone must still be marketable to the public and profitable to the maker
it must be safe and long-term reliable

Here’s what AI needs to do:

Find its own information or data which it recognizes as relevant to this problem.

Know the laws and principles of chemistry, physics, materials science and electrical engineering.

Recognize or reasonably deduce relevant manufacturing methods.

Know something about the business model for cellphones.

Go through scenarios and find that battery type.

It might also think outside my question and…

look at ways to reduce energy consumption which might make the current battery types plausible,

design new operating software which might require less energy,

design new hardware which might require less energy.

I am actually interacting with folks in the AI development and implementation space. Thus far my experience is that most of these things that AI needs to do still requires human intervention and I haven’t seen an amazing invention yet come out of AI.

So far, most of what I see is AI being used to write articles, replace customer service and make entertainment. Frankly, the majority of us still recognize if something is AI vs. the real deal. So, it’s not there yet even with tasks simpler than the cell phone request. Frankly, people are still needed to correct errors. Often human bias is obvious in the results.

Maybe it has the potential to go after these more complex tasks and there is still hope that many beneficial advancements might come in medicine, communications, energy, transportation, materials, environmetnal stewardship, etc., etc. But when will that occur? In the meantime, existing software, hardware, programming methods and supercomputers can continue doing what they do (as can the humans). But it seems that all the investment is in anything thst says its AI. In fact, lots of needed technology areas are being avoided by investors, to civilization’s detriment, to chase AI.

In 1929, the Marx Brothers generated a talkie called “The Cocoanuts” about the Florida real estate boom and bust of the 1920s mixed with other assorted cons and frauds. Fast forward to the 2020s and The Cocoanuts, it turns out, are still dropping from the trees. I’m convinced, based on my own review, that at least 3/4 of anything called AI is just what has been done before (as I described) dressed in AI clothing. Of the other less than 1/4, there may be legitimate attempts to foster AI towards the singularity (mimickng human consciousness) and/or to use AI agents to solve a problem or implement a solution or modify something to make it faster, lower cost, solve a problem, etc.

Does this require the enormous number of AI data centers which are plastering the country and seem to consume enormous amounts of resources? Yes and No.

Yes, for the following reasons:

if you are interested in replacing humans with AI agents for many business and government tasks
if you are interested in surveying human activity which requires enormous processing power

I understand the desire to replace humans in many tasks. The AI agents don’t complain, show up late, get sick, need workmen’s compensation, need health insurance, get paid (although the AI technology and data center companies do) or sue. This has been a long-term goal of many corporations and shades of this have been around for years. Ever try to get a human on a customer service call? However, at this time, I wouldn’t trust AI to do much in the science or engineering space that might directly impact people; think brain surgery or bridge design as examples. We’re also a long way off from changing my car oil and fixing a toilet.

The second one is concerning as I am a huge advocate of freedom and privacy. Proponents argue they need it for safety and for marketing. You can dwell on that.

No, for the following reasons

if the goal is making new and beneficial technologies (there’s still lots of money to be made here)

Frankly, most AI data centers are not needed for this activity. This can be accomplished with 1/10 of what’s built and planned but this is not the only driver for AI data centers.

Robot Image from Pixabay

Before I end this article, which won’t be the last on AI, consider that robots and AI have become intertwined in the public imagination. Humans are rightly concerned about the idea of robots becoming conscious and thinking for themselves. After all, if they think like us that might not be a good thing. If they believe they are the master-species, watch out. Just keep in mind that they require a power source and batteries don’t give them a lot of up time without swapping for a new one or re-charging. A robot plugged into a power source doesn’t have much ability to move very far or for long time periods. I can think of ways to power robots remotely but I’m sure other people are thinking of that. In the meantime, maybe steer clear of the Cocoanuts, they hit hard when they fall.