In Jewish mythology, a golem is a giant creature made of clay. It is inert until awakened with the enscribing of a secret word. It has no will of its own. But with a single command it can, and will, destroy whole cities. Remove the command and it is powerless. Alas, the same cannot be said of the modern golem.
The modern day golem is far more insidious and the destruction it brings is not to cities but to truth. GLLMM stands for Generative (self-taught), Large Language (responding to commands in plain language), Multi-Modal Model (treating text, video, and sound the same).
The term was coined by computer scientists Tristan Harris and Aza Raskin from the Center for Humane Technology. These “golems” are Artificial Intelligence engines that have been trained on everything you can find on the internet.
The big five technology companies, among others, are all racing to create the most powerful, most user-friendly generative AI while at the same time crying out for regulation because they know how dangerous this is. ChatGPT, Google Bard, Microsoft Bing are just a few of the Generative AI that have been released.
The good news is that these golems may help us cure cancer and survive climate change! They are so adaptable that many businesses will be able to run without any humans at all! Film studios can replace actors with AI generated characters. Publishing houses can replace authors. Students can generate term papers in an instant.
The bad news is that you need no special knowledge to use generative AI. Anyone with a grievance can now create “deep fakes,” mimic anyone’s voice, create perfect false recordings, or write “scientific” papers. Generative AI has no sense of right or wrong. It does not care if it is trained on truth or lies. And they are amazingly good at doing anything anyone asks them to do.
The world is full of good, caring, sensible people. It is also full of con men, disaffected youths, vengeful exes, and extremists willing to do anything to achieve their goals. Oh, and dental plan companies (just ask Tom Hanks).
It does not take a creative genius to imagine the harm they can do.
The one good thing that may emerge from this experiment with nearly-conscious computers is that people will finally realize that you cannot believe anything you read, see, or hear on the internet.
We need to hold actual human beings responsible for unleashing these monsters. The danger comes, not from the AI itself, but from the ease of abuse they offer. The programs are already out there and, sadly, the generative AI designers failed to include an off switch. This is the AI apocalypse we should all be worried about.
Lisa Wright divides her time between her home in Lakeville and Oblong Books in Millerton, where she has worked for nearly 40 years. Email her at wrightales@gmail.com
