Turing

AI is used to write a TV show. Billed as a "Deep Blue" and "Watson" style challenge.

Notes

Characters

• Owner of Netflix style movie service
• Stan Ulam: Elon Musk style investor
• Freddie Williams: Created the core algorithm for his Ph.D at Carnagie Mellon

Plot Elements

Conflict with writers union. This will end up tying back around as the AI being a tool. Cite the chess games after Deep Blue. Also hint of it being used in writing political speeches. And tailoring question and closing arguments for lawyers.

The show is a sci-fi show and ends up having the "robot betrays inventor" trope. Themes on children, ties back to Frankenstein.

One of the episodes ends up getting a really bad review and there is a lot of debate over where the blame lies. The fact that it's not clearly with the script ends up being a strong argument for the AI.

Philosophical Elements

A machine is just as "happy" doing what it was designed to do as a person is doing what they were "designed" to do.

We don't lose control of our creations often, instead we lose control of people.

What is creativity? Is a person remixing the cultural elements they've learned more authentic than a machine doing it.

The Musk character is seen by many as the face and therefor the creator. But as the project is unveiled to the public, he is adamant that he didn't build it, just saw the value and funded it. Focus shifts to Fred Williams, and he tries to explain that he developed the core algorithm, but it was only a variation on earlier algorithms and the real work was in tuning it and building the modules.

Story

Freddie Williams sat down in front of a panel of six. Five of the members were hard scientists. Or mathematicians. That was an eternal debate amongst them. But they all agreed the sixth member did not belong because he was the "old kook". Not that they would ever call him that outside their most guarded moments, calling Tom Kilburn, one of the most eminent philosophers, an old kook would be career suicide, even for an esteem professor of computer science. Even if all reputable computer scientists knew he was an old kook, it simply wasn't said.

Before any of the panel had a chance to begin the inquiry, Freddie jumped into his defense. "I'd like to start with two confessions. The thesis (thesi?) you have in front of you were not penned by hand. I was aided by a machine, I call it a 'printer'".

Five of the six member were not amused. One was grinning and after jotting a few notes, started twirling a pen playfully.

"The second confession is that the words were not penned by my mind. I was aided by a machine, I call it 'Turing'".

One of scientists leaned forward and asked, "I'm not sure I follow, your thesis is about 'Turing'. You have examples of nursery rhymes and children's stories it composed."

"Well, I started there. I fed it a corpus of every children's story I could find. And it produced stories that had plot, characters, even the prose was not bad. But they didn't fit together right. Most children's stories have a moral. Some point that the story is getting across. Without that it's just sort of a fever dream of princesses, dragons, knights and talking cutlery. That was the real challenge, giving Turing a goal - make this point. The essence of that algorithm - well, it's five algorithms and two data structures, are described in detail from pages 72 through 112. The next part isn't in the thesis though.

Once I had that, it turned out to be flexible. One time I gave it the instruction to produce a story whose moral was 'Always try a Monte Carlo simulation first', I was a bit frustrated at the time. But to my surprise, it used it's web crawler to pull in enough information to describe a Monte Carlo simulation. Sure, it was a talking badger explaining Monte Carlo to his friends and how his uncle used it to steal the kingdom that was his hereditary right. But it wasn't a bad summary. So I switch the corpus over thesuses (thesi? please somebody tell me what word to use here) and with a little tweaking I was producing pretty decent papers.

The last step was to convert the elements of my research over to a story it could tell. The result is the document you have in front of you. And three other dissertations that I sold to panicked grad students. Kidding."

The level of wispering and note passing among the five very serious members of the committee had been escalating as Freddie explained the source of his disertation. Now that he sat silently it rose to a very quiet fervor. Before they could ask any of their questions, Tom spoke up.

"Is Turing conscious?", asked the old kook.

"Sort of. Considering he wrote a very technical description of himself, I would argue he is self-aware. He is capable of abstract thought, even directed self learning."

"I'm not sure the term 'sort of' belongs in a rigorous defense of a computer science thesis", said the old kook as he propped up a straw man, then leaned back in his chair to show Freddie he was going to stay out of his way while he skewered it through the heart. I can't Fur Elise (at least not past the first few bars) so I can sort of play. And .

"Sure it does. When I get a 76% on a math test, I sort of understand that subject. If Turing meets 76% of the definitions of conscious, he's sort of conscious. I'm awake and talking to you, so I'm sort of conscious. I'll be less conscious tonight when I'm sleeping, but more conscious this afternoon when my hangover wears off"

The video of Freddie sharing a smile with time froze as Stan Ulam enthusiastically tapped the spacebar on a thin computer.

"We hired him shortly after that. Actually, you can see me right there in the corner with my phone. I was sending a message that we had to hire him then. He's continued working on Turing, he's got a whole team now. But we have a perception problem. We need a demonstration of the algorithms power - Deep Blue vs Kasperov, Watson on Jeopardy. We want to use Turing to write a show."