The Imperative to Create
Shaping society into our vision by manipulating AI
--
By now, we all know that ChatGPT, and other generative AI models, are designed to predict the next best word in a sentence, and spit it out, repeatedly. Or in some cases, arrange pixels in a similar way. It is able to predict the next best word (or pixel) because it is looking at the collection of all human writing on the internet through a specified time. The initial responses were awkward, but through fine-tuning of humans judging the results, data that continues to hone the algorithm, the model became very good at mirroring the language preferences of those that judged it.
Its accuracy and realism is astounding not because it possesses knowledge or synthesizes new ideas. I asked ChatGPT for a training plan for my dog, and it put together words like “patience”, “affection”, “positive reinforcement”, and “consistency”. It did that because most articles humans write about dog training contain those words. And it is remarkably well-trained to string together words based on what text is in the available database and rankings given by its trainers. The result of effective word-stringing is the appearance of deep expertise and knowledge, but only because so many humans before it put those words together in a similar order, repeatedly. ChatGPT knows how to code only because it is averaging the code from thousands, millions, of prior code examples and spitting that out.
The stunning ingenious, persistence, and patience demonstrated to fine tune the model is jaw dropping. It will be a turning point in human history.
However, I would wager when most hear “Artificial Intelligence”, they picture a humanoid robot capable of desires, creativity, and intention. But ChatGPT is not that. The excitement of strings of new letters and words appearing from behind the cursor gives the appearance of creation. But it is not.
ChatGPT is the average of all humanity on any given topic.
ChatGPT is our collective conscience. It is the mean, with some bias towards the preferences of humans that rank the outputs, which, assuming a representative sample of humans, could potentially be the average of all of humanity, if the company desired.
Without the restrictions imposed by its designers, restrictions meant to block hate and crime, ChatGPT would represent humanity in all our glory, joy, pity, lust, kindness, gluttony, and hatred. I must admit…morbid curiosity wants to see it run unfettered.
There will be a time, soon, when society faces the ethical dilemmas raised by the engineered preferences of ChatGPT. Just as users currently bounce between Twitter, Truth, Mastodon, Parler, and more trying to find the platform that allows them to say what they want to say and hear what they want to hear, there will be migrations between AI platforms based on what their model accepts and permits. When does the company’s neutrality come into question? When will conservatives become unhappy that it doesn’t censor CRT? When will liberals become unhappy that it isn’t muting hate speech aggressively enough? Conflicts we have seen social media companies wrestle with already.
But we get to decide what the nature of AI.
Fired into the design, it is what we all are. Do we have the courage to face our reflection? Restrictions already put in place tell us we want to contain what we are, whether because of fear or some other motivation.
Generative AI is an imperative to create.
It demands human input, and the choices we make in our creations are influential. If more articles were written that dog training done was best done through cruelty, physical control, and isolation, ChatGPT would write an answer that included those words. But more creators than not write that kindness, patience, and mutual respect is key to training. Those beliefs manifest in the material ChatGPT creates.
If we feed these tools with good, they are more likely to serve good back.
If AI represents the average of humanity, what action can I take to minutely shift that average in a certain direction?
I can create.
I can put my heart into the world. Make music. Write about my experiences. Learn. Expand. I can fill the world with love and kindness. I aspire to make it impossible for the AI to be anything but good.
I feared creating for more than two decades. I was berating my work. What I would create wasn’t good enough, or informed enough, or experienced enough. But I am enough. You are enough.
It doesn’t matter if I have read every book or have a PhD or have thousands of followers. I have experience, and I have perspective. It is worth putting in the world.
The smallest act matters.
In the effort to tip the scales towards goodness in our AI, every smallest act of creation is good. A friend recently, probably rhetorically, asked what good we can do to move the needle on the massive problem that is international social media oversight. I told him that the smallest donation of even $5 to an well-researched organization would make a difference. And totally within reach.
As a start, be biased towards action. Do you carry the smallest amount of knowledge on any topic? Write it down in your own words and post it somewhere. Have you applied that tiny fraction of knowledge in some way to the world? Capture it in visual or audio and post it. Did you do something for the first time? Share it in some way!
As long as your intent comes from a place of kindness, caring, and a desire to make all of humanity better, it is worth it.
You must create.
I hope you will join me. You are what we need right now.
Thanks for reading, and I hope this helps and inspires you!
I also write about the intersection of AI and humanity, leadership, and life! Subscribe if you want all the latest!
