Dr. Henry Frankenstein turned to his associate, Victor Moritz, and yelled “it’s alive!”
The doctor didn’t say “she’s alive,” “he’s alive,” or “they’re alive!” He said: “Look! It’s moving. It’s alive. It’s alive… It’s alive, it’s moving, it’s alive, it’s alive, it’s alive, it’s alive, it’s alive!”
The word of the day is “personification.”
- Football is the King of Fall (238 words)
- Breaking Alive (219 words)
- Dungeons & Damages (333 words)
Football is the King of Fall
When I was 7, I saw a poster with the word “personification” printed on the top in bold letters above a football with eyes, a mouth, and a crown on its head. Beneath the image, were the words “Football is the King of Fall.”
Being autistic, it always struck me as odd. Does everyone like football? Aren’t there people who prefer fall for the new TV series, the fact that it’s the start of hunting season, the end of the MLB season, or any number of other fall-specific events?
Over time, of course, I realized the point of the poster wasn’t to express an opinion, but to demonstrate what “personification” was. We see and hear it all the time. A lot of people still call ships, cars, and aircraft “she” despite the clear misogyny.
And, now, you can’t throw a rock without, for example, someone giving their “Replika” a gender designation. This morning alone I saw three different STEM workers describe ChatGPT as “he” or “him.”
When you consider that the average person knows about as much about AI technology as they do dark matter or fusion research, this is a recipe for grand-scale ignorance.
Everybody knows that footballs don’t wear crowns and declare themselves seasonal monarchs. I don’t think the same general intuition grips the public when it comes to personifying AI technologies. I think this could pose a serious problem for humans and machines in the future.
Could a Replika bot become sentient, choose a gender, and manifest as an efficacious human-adjacent entity deserving of its own self-wrought identity?
Who cares? Anything is possible in an infinite universe, but you can’t measure “potential for sentience.” And if there’s nothing to measure, you can’t do science. And, if you can’t do science, all you can do is guess.
I guess the sun is God because I can’t figure out what it’s made of and it keeps me warm. I guess the Earth is flat because I see a line on the horizon and I lack the technology to explore it. I guess AI could be sentient because I read a science fiction book about it that made sense to me.
Without something to measure, I can’t do science. That’s why I can’t tell you if rocks are sentient. Nobody can. They exhibit absolutely no evidence of sentience, yet I cannot dismiss their sentience unless I can quantify potential.
This is a direct extrapolation of Heisenberg’s Uncertainty Principle, that states: the position and the velocity of an object cannot both be measured exactly, at the same time, even in theory.
Heisenberg couldn’t measure potential from a snapshot, and neither can we.
For this reason, anyone who claims to have a benchmark for sentience is almost certainly being disingenuous.
Dungeons and Damages
Despite what you may think, I absolutely love AI. I think LLMs are among the most important inventions in human history. Simply put: they allow us to communicate with each other in ways our predecessors could have only dreamed of.
Imagine, for example, having Facebook Translate services available prior to the onset of World War II when the Japanese Empire was trying to get a seat at the negotiating table.
It’s an understatement to say that LLMs will change everything. They’ll democratize the expression of imagination. People who lack the talent of their favorite writers will finally have a writing assistant that can help them with “voice.” Those who can’t visualize art from text prompts or “hear” instruments as they read music notation will finally have the opportunity to manifest their mental states as novel art.
Ray Bradbury’s “Harrison Bergeron” warns us of the potential dangers of such a state — when everyone is made equal, nobody can be allowed to become exceptional — but I think the reality will be much, much more positive.
For example, as an experienced Dungeons and Dragons player who understands how generative models such as ChatGPT work, I know that in 2023 I can pretty much show up to game night with no prior prep work done and ‘dungeon master’ (be the person in charge of) a full session as long as I have internet access.
That’s something I couldn’t do in the beginning of 2022. This is going to change the way I dungeon master.
And, once everyone else figures out that you can get these models to generate infinite rules, lore, and art, I imagine we’re going to see massive upheaval in the tabletop roleplaying game (TTRPG) industry.
Heck, with the right prompting, you can get most of the popular models to output actual text from beloved TTRPG titles from numerous publishers. Remember, these models were trained on the internet.
One imagines that the Kevin Siembiedas of the TTRPG world aren’t going to take this well.