"How do you know about all this AI stuff?"
I just read tweets, buddy.
Page 7 of 37
ChatGPT loves to pick 42 as a random number. Of course GPT-4 can run some Python code to correct it, this could help some folks think about the non-random nature of things they assume might be random when they ask GPT to "choose."
It's kind of obvious that most people involved in this AI hype-cycle today are excited because they assume continuous "infinite" improvements over time. But what they seem to fail to take into account is the embedded error rates (as these are heuristic systems by definition) that you'll never be able to get rid of.
...
We may very well get a decent AI based text summarization tool, but it will certainly have to use a different model than current architectures imo.
Love it, filing under #lol
for pain.
These are honestly not that interesting.
I think red teaming by people who want the system to fail is probably your best bet. Get some real haters on board.