I understand we have always conventionally transported goods by horse. Yes, this employee knows nothing about horses, and in fact is rather spooked by them, but we've checked! Their claim to be able to transport goods faster, without a horse, somehow seems to hold up.
Maybe, just maybe, we should take this whole "truck" concept seriously?
Humans can only usefully steer LLMs if they have some understanding or context the LLMs do not.
It's like the /s sign. Can some people not, for themselves, realize that this text is meant as a more or less a joke ? And before you ask, yeah, I am aware some of the people are on the spectrum, but still...
Is humor that hard to grasp on the internet ?
Guess that makes it official then. We are all automatons pretending to be flesh :-)
* some sincere people with very extreme takes, * some trolls that masquerade as the above, to bully others over their credulousness and lack of guile, which is distinct from sarcasm, * some trolls that insincerely speak anything that earns engagement, * and more and more bots that mimic the above
So sadly, the answer to your question is generally yes.
What I'm curious about is this: in the end, experts are the ones who are best at distinguishing hallucinations. If you can just search with GPT and tell the difference, wouldn't that be enough? I can't imagine memorizing thousands or tens of thousands of lines.
I have ADHD, and when I get nervous, I tend to forget what I was going to say, so it's even more true for me.
A qualifed researcher would have had their agent perform the talk on their behalf rather than waste everyone's time.
At the current time I'm writing this, all other top-level comments are engaging with the article as if it were sincere. So, yes.
topham•56m ago
So, plagiarism. Daily.
TheJCDenton•53m ago
rolph•44m ago
bena•18m ago
wrs•7m ago
handoflixue•45m ago
I don't get how you can possibly call it plagiarism if it produce a novel breakthrough - by definition, the existing knowledge base doesn't contain the new ideas generated in this process.
And we've proven it can handle complex, novel thinking when it solved a significant Erdos problem back in May: https://www.scientificamerican.com/article/ai-just-solved-an...