Artificial intelligence is upon us. Either a nightmare in the making or a transformational evolution between man and machine that could free us from the mundane tasks of the world. The problem is, that those mundane tasks are all being handled by a human who is going to lose their job. A semi-educated human with general skills and very little specialization. The “C” student. Those jobs are now at risk. But this isn’t new.

Globalization

Remember The World is Flat by Thomas Friedman? He predicted that global high-speed internet and low-cost labor were going to do away with menial jobs. Hasn’t happened yet. He cites a study in his book where a McDonald’s franchise outsourced the drive-through order taking. The order takers in Bangalore, India were faster and more accurate. Did McDonald’s suddenly start offshoring drive-through tellers? No. Why not? What happened? Did Friedman miss something?

Faster & Unlimited Power for AI

Today’s AI is a bit of a mess. We are trying to understand if the LLMs are going to kill us or make us millionaires. In terms of intelligence, AIs are really more like massive predictive databases that use math to organize words in patterns that mimic previous groupings of words. We’re in the phase of “what’s the next right word” intelligence. There’s a big element missing. The human element. The indescribable human. The uncloneable brain. The leaps and skips of an engaged mind. It’s okay, AI is going to get better and faster, but at this moment, it is not clear how faster or smarter chips are going to give AI a soul.

That’s where “training” comes in. Humans now augment AI. It used to be the other way around. And, I suppose that today, the ChatGPTs and GPT-of-the-week, do appear intelligent and useful. They are not very creative. Even after training, ChatGPT 3.5 can’t write a decent free verse poem. It can mimic a poem. It can demonstrate that it understands “free verse” to mean non-rhyming, and then spit out another rhyming verse. It’s not very artistic. It doesn’t have the ability to see outside the box of letters it contains to contemplate the nuance of human intelligence. There is no intelligence yet. That’s where the humans come in.

Today, humans (you and I) are training LLMs to be better mimics. Faster mimics. Does adding 100 Nvidia processors to your AI LLM give it wings? No. It just runs faster. And with the upcoming ChatGPT 5.o, perhaps closer to creativity. But that’s really where humans have the edge. We can direct AI to create a drawing of a mountain range. We can redirect the AI over and over, refining and retraining the LLM for what we hope to create. There’s an interesting experience when art directing dall-e for example, of having a willing artist following your suggestions and requests over and over. And… still not getting what you want. Not exactly. Here are a few examples of my dall-e chat. I started off with a red crayon direction but ended with sumi brush. None of the results were what I was looking for. I continued to “refine” my prompts. And I continue to get poor results.

.

Then, I just Googled “Japanese style sumi mountain lake ink painting winter no sun” and I got close. And I also got a lot of clipart of “Japanese style sumi-e paintings.” And “Millions Of Art Prints.” Oh shit. Do you see where this is going?

Now that ai’s are free and art is an objective term, be prepared for the “art” to be flooded with mimics and prompt-art. It’s not art, folks. It’s not human. It’s ai-art.

What About Business?

AI today offers huge opportunities for business processes. Old math-ish skills like SEO and ad optimization is going to be picked up by ai tools and those jobs are going to be more about training and monitoring the ai’s projects. We’re going to be project managers for our ai copilots. While that’s a useful and important skill, the prospect of trying to teach ChatGPT the difference between “free verse” or “sumi brush paintings” does not sound like a joy. Where is the joy in ai?

Creative directors will still be in dire need. Maybe even more so. But, my experience cojoling dall-e to give me a usable image was not all that pleasant. Once the “magic” of ai generative art wears off, we’re going to hit the hangover phase of ai. With all this generative content, where’s the humanness? What is the soul of a sumi brush painting by an ai?

Like Microsoft’s new “copilot” feature and Google’s “assistant,” we’re going to have a lot of opportunities to interact with and “train” ai. What we’ve got to figure out is not if the algorithm is going to kill us, but can we make it creative? Can we wake up the soul of ai?

Today, I’m thinking even a quantum leap in processing power and available LLMs is not going to give ai-art a creative leap. That spark we’re all anticipating, the general awareness of ai, is a long time coming. Perhaps a breakthrough in some unforeseen creative branch will spark a new way of organizing and sorting the incoming data into something resembling creative ideation. I’m anticipating that moment. I’d even work to train that ai, when it arrives.

I am not, however, holding my breath. Here is what I had in mind.