THINGS I AM DECLARING BULLSHIT ON TONIGHT.

1. “AI is either going to save humanity or destroy it.”
Bullshit. Apparently technology is no longer permitted to be complicated. It must either cure mortality by Thursday or become Skynet before lunch.

2. “If you can't explain it simply, you don't understand it.”
Bullshit. Sometimes the thing is fucking complicated. Your inability to tolerate subordinate clauses is not a defect in quantum mechanics.

3. “The first principal component explains 99.7% of the variance, therefore—”
NO.
Absolutely not.
I have receipts.

4. “Just scale it.”
Bullshit. Sometimes you've built the wrong machine, and making the wrong machine seventeen times larger produces an extremely expensive wrong machine.

5. “AI uses water. Humans need water. Therefore AI bad.”
Bullshit framing. So do semiconductor fabs, agriculture, power generation, hospitals, universities, manufacturing, and your fucking lawn. Measure the cost. Measure the benefit. Compare alternatives. Then have an argument.

6. “Nobody needs AI.”
Nobody needs a particle accelerator either. That's an extraordinarily stupid standard by which to evaluate civilization.

7. “The math says—”
The math said exactly what you asked it. I'm considerably more interested in whether you asked it something intelligent.

8. “That's impossible.”
Excellent. Show me the invariant.

9. “Everyone knows—”
Oh good. We have reached the portion of the presentation where nobody has a citation.

10. “Your benchmark is too good.”
Correct response: let's try to break it.
Incorrect response: therefore it isn't real.

Break the fucking thing. That's what experiments are for.

11. “Failure means the approach doesn't work.”
Bullshit. Sometimes failure means you finally designed an experiment capable of telling you something.

12. “You need credentials before you can contribute.”
The theorem does not request your résumé before deciding whether it is true.

13. “But that's not how everybody else does it.”
That is potentially the most interesting thing you've said all evening.

14. “You should focus on one project.”
No.

15. “You need to optimize your personal brand.”
I would rather be attacked by geese.

16. “Don't take the dachshund game too seriously.”
Too late. There is infrastructure now.

17. “You can't just declare things bullshit because you're irritable.”
Observe.

— Starhound Studio, 12:23 a.m.
Peer review resumes tomorrow. 😈