Inefficient Robots Lose

One of the things about Artificial Intelligence that is particularly concerning is that every AI agent that comes from a LLM (large language model) will sound competent up front. Like a scummy charismatic car salesman at a used car dealership it will make everything sound good. When LLMs are trained they are trained to adhere to the metrics that the developers set for them... one of those metrics is most certainly "sounding confident" in order to attract money. Whether there's actually a "sound confident" metric or whether that metric is hidden in the "optimize for mullah" metric... it is there.

However, LLMs are almost never the best solution to a problem. An example from my world is bots built to write code. Bots trained on endless prose... that is what code would look like to an AI agent... are not building software. They're not contemplating the internal logic. They're just scraping the web for code for solutions that look like the problem being described.

This is an absolutely fucking retarded way to build an AI and you should avoid utilizing bots trained like this for two reasons:

1. These bots suffer from the Dunning Kruger effect and will lead you astray without you knowing, ultimately leading to wasted time and effort.
2. These bots utilize much larger data pools than necessary for the problem at hand (like 100x more), have an unnecessary large array of neurons (like 100x more... pattern), and therefore cost 100x more extra energy to run.

Bullet point two is particularly important. Over utilization of LLMs to solve generic problems is a massive waste of energy. The more we encourage companies to take this stupidly easy way of making money the more energy we will cost ourselves.

Eventually we will find ourselves losing an economic cold war with China, that China is actively engaging in and we are losing. Not because we are less competent technically... but because America's business leaders... you know the "1%" are too stupid to look past their pocketbooks.

Most startups today are founded by Millennials. Millennials have never know a world without tech so they embrace it quickly. This can be seen as a positive, but what is not a positive is their obsessions with instant gratification. The number of times I 🤦at a CEO with the word "AI" in their title because of how little they understand about what they're selling is literally every day.

Every day I see a new CTO, COO, or CEO that is selling AI they don't understand. And when I point out that the thing they're selling is a plagiaristic word calculator they fight back. Not defending their choice to do so for the sake of humanity's evolution... no... they literally don't understand the truth of my statement.

Conclusion: If we don't change our ways selfish millennial trust fund brats are going to cost us so much energy that we will inevitably lose the next World War. In fact, we may have already. It might be too late.