I definitely agree that ChatGPT can be incorrect. I’ve seen that myself. In my experience, though, it’s more often right than wrong.
So when you say “in nearly every question on complex topics", I’m curious what specific examples you’re seeing.
Would you be open to sharing a concrete example?
Specifically: the question you asked, the part of the answer you know is wrong, and what the correct answer should be.
I have a hypothesis (not a claim) that some of these failures you are seeing might be prompt-sensitive, and I’d be curious to try it as a small experiment if you’re willing.
This is basically the solution. Light travels about 1 foot in one nanosecond, so the car needs to reject latent replies.
I did research in this area a few years ago. Here's a research paper [1] from 1993 that goes into more detail about this type of "distance bounding" solution (i.e. authenticating received signal only if 1) it is received within a few nanoseconds AND 2) the decrypted received signal contains the previously sent random number) in order to defend against "relay attacks". The paper discloses many variations to this general solution as well.
[1] Brands and Chaum, "Distance-Bounding Protocols"
For me, I use my annual checkup to get data (i.e. lipid and comp panels, etc, hearing test, electrocardiogram, urinalysis) on myself so I have a baseline of what the data looks like when I'm healthy. Everyone has slightly different baselines, so it's nice to have a better picture/time-series of mine.
You make a good point about ease-of-use. I agree a phone app is much easier to use with a smartphone. However, people with flip phones couldn't install such an app. You might then argue the demographic with flip phones would either use an RSA device or not have 2FA enabled at all - which seems like a valid point.
Security-wise, having a secret user math function seems more secure than the Google app. I can give reasons why if needed.
Edit: I had an idea for an improved sms 2fa, but comments gave persuasive reasons why google authenticator was better. Thanks for the comments!
Idea basically is a 3FA system where bank sends you a one-time 6-digit number. You then have to translate that number using a user-seeded cryptographic hash function. This secret function is your third factor which translates the received SMS code into the value you'll input at login.
Analysis: Security would increase; but ease-of-use would decrease, especially in regards to how a user would reset their password if they lose both their password and their program that calculates the cryptographic hash.
Thanks for this insight! I'll edit my comment to credit you, but I won't delete it since someone might have the same thought process as me.
My comment:
So I see now (thanks to you) that it is just showing test cases (test warbles) to demonstrate that these scrambling techniques work with foreign languages. However, why would the us gov need to make sure that this program can successfully obfuscate Unicode strings in Chinese/Russian/Arabic/Farsi?
My gut reaction: while code comments would be trivial to forge, it appears the us gov is still using foreign language strings in some way - maybe having just one string constant originally in a foreign language that is then obfuscated/scrambled (such as by xoring every char against a random key)
Edit: part of my comment is corrected by comment below - Thanks openasocket!
Another comment about the content of this article:
Three quarters down the wiki page there is code for "adding foreign language" to the code. The options are are to add code comments in Arabic/Chinese/Russian/Korean/Farsi. My gut reaction is the purpose of this added language is to obfuscate the true source of the code - i.e. the code has Chinese comments in it so it must be from China. Ahh. I guess this makes sense to do. Only problem now is that the Chinese/Russian/Farsi/etc characters that they included in their code is now public. (Obviously now the CIA will change the foreign language words they insert)
I'd posit if someone had an X-year-old (i.e. x=7) copy of some malware, and the malware had these specific foreign language comments as shown by the article, there's a good possibility the source of the malware would be from the us government.
Hi, it looks there is a factual dispute about the linked article. I think I might be able to add some value to this conversation (but that's of course for you to decide).
It appears the parent poster is arguing Google did ban [all payday loans] while you are arguing that the article says Google did not ban [all payday loans], but instead only banned [loans with interest rates >=36%].
My viewpoint: The article says both points, i.e. Google banned [all payday loans] and Google also banned [all loans (i.e. of the non-payday variety) with interest rates >=36%]
Evidence / Recitation from article:
"...In addition to the broad payday loan ad ban, Google will not display ads from lenders who charge annual interest rates of 36 percent or more in the United States. The same standards will apply to sites that serve as middlemen who connect distressed borrowers to those lenders..."
"...Google announced Wednesday that it will ban all payday loan ads from its site..."
As seen from the first above recitation, the words "in addition to" appears to mean that two separate bans have been enacted: The first ban is for any loan classified as a payday loan. That means a payday loan of any interest rate (i.e. 35%, 25%, even 3%) will be banned. The second ban is for a loan of any type where the interest rate is >= 36%.
From reading the above report, it seems it was officially only a one-lane road, but the road was big enough to handle two-streams of traffic. The google car could have just stayed in the middle of the road, but instead, was hugging the right side of the road in preparation for a right-hand turn. Due to sand bags next to a storm drain, the google car had to "merge" back into the one-land road to get around the sandbags. Considering it's still a one-land road, the bus driver should have yielded to any car that was in front of it. I'd place a majority of the blame on the bus.
What I don't understand: Why doesn't the Google car have video of the accident? Or if they do, does anyone know if they will share a video of it?
So when you say “in nearly every question on complex topics", I’m curious what specific examples you’re seeing.
Would you be open to sharing a concrete example?
Specifically: the question you asked, the part of the answer you know is wrong, and what the correct answer should be.
I have a hypothesis (not a claim) that some of these failures you are seeing might be prompt-sensitive, and I’d be curious to try it as a small experiment if you’re willing.