Trick question. Humans are biological machines.
Randall Jarrell said a fellow poet’s works sounded “as if they were written on a typewriter by a typewriter”.
I wonder what he’d say now.
6 for 6. The algorithms are following rules for rhyme and structure, but aren’t conceptually consistent, whereas the humans are clearly trying to have an underlying conceptual thread from start to finish.
A group of 10 judges weren’t fooled
This seems to be central to many people’s definitions and benchmarks about AI. I think it is far more telling that humans are easily impressed by a capacity for deception. This capacity evolved in organism for specific biological reasons, and arguably has no function in non-organisms, who don’t need to reproduce, maintain life processes, nor be concerned with localized resources.
Just like there would be no benefit to meeting aliens who perfectly replicated human physiology and behaviors, I would rather value what artificial life and intelligences bring as their own strengths and qualities. The poetry that a data system writes for itself or other such systems is more interesting than the naive egotism of getting it to mimic humans.
Yes but can they rap?
Hell, it’s about time someone told about my friend EPICAC. After all, he cost the taxpayers $776,434,927.54. They have a right to know about him, picking up a check like that. EPICAC got a big send off in the papers when Dr. Ormand von Kleigstadt designed him for the Government people. Since then, there hasn’t been a peep about him–not a peep. It isn’t any military secret about what happened to EPICAC, although the Brass has been acting as though it were. The story is embarrassing, that’s all. After all that money, EPICAC didn’t work out the way he was supposed to.
And that’s another thing: I want to vindicate EPICAC. Maybe he didn’t do what the Brass wanted him to, but that doesn’t mean he wasn’t noble and great and brilliant. He was all of those things. The best friend I ever had, God rest his soul.
You can call him a machine if you want to. He looked like a machine, but he was a whole lot less like a machine than plenty of people I could name.
Agreed. The Turing test (and derivatives such as this poetry competition) are all very well, but maybe AI that can fool humans isn’t a helpful goal or even a very good idea. I suspect you’ll agree that the Winograd Schema Challenge makes a better benchmark for real-world intelligence.
Also, 6/6. Like @tropo says the machine poems clearly aren’t about anything, the humans’ are.
Turing doesn’t test for artificial intelligence, it tests for virtual intelligence.
And, yeah, not terribly useful.
This topic was automatically closed after 5 days. New replies are no longer allowed.
