Hacker News new | past | comments | ask | show | jobs | submit | drew-y's comments login

I wonder if we could add some type of verification registry. It would be nice if browser's could have a big indicator saying that this website is verified to associated with Dell inc.


Some HTTP certificates do exactly that, and web browsers used to show the company/identity the certificate was issued to in the URL bar. Now you have to go to the certificates detail, very clear on Firefox, behind a few clicks on Chrome. Here's an example from a bank in Spain: https://www.bbva.es


HTTPS certificates should do exactly this.


They should. And sort of already do. Though, I wonder how difficult it is to register with some certificate issuers under a fraudulent name.


That was EV certificates. They were finally removed from browsers completely around five years ago because they didn’t actually work. At all. The problems were largely social. Plenty has been written about it, you can find it by searching.


Well, the original HTTPS certificates too were supposed to work like that; I remember reading a security article criticizing the EV proposal by quoting the old (circa 1998?) policy statements of different CA's and showing that they're pretty much identical to the EV requirements.


How does automation reasoning actually check a response against the set of rules without using ML? Wouldn't it still need a language model to compare the response to the rule?


aiui a natural language question e.g. "What is the refund policy?" gets matched against formalized contracts, and the relevant bit of the contract gets translated into natural language deterministically. At least this is the way I'd do it, but not sure how it actually works


Awesome Job! The animations on the characters are really great and do a lot to make the game feel more alive.

Here's some ideas for improvements you could make:

- Add a button to go back to the main menu / exit the battle, I keep hitting the home button instead

- Add some sort of indicator to make it clear when an attack is charged. Also maybe some hints as to how attacks are charged in the first place.

- Consider moving the question input to inside the game instead of the alert

- Try playing around with the colors to make text easier to read.


Thanks for checking out the game and for the feedback!


Totally agree with the sentiment. I find the constant debates on "is AI conscious" or "can AI understand" exhausting. You can't have a sound argument when neither party even agrees on a concrete definition of consciousness or understanding.

Regarding this line:

> ChatGPT can already tell you a lot about itself (showing awareness) and will gladly walk you through its “thinking” if you ask politely.

Is it actually walking you through its thinking? Or is it walking you through an imagined line of thinking?

Regardless, your main point still stands. That a program doesn't think the same way a human does, doesn't mean it isn't "thinking".


> Is it actually walking you through its thinking? Or is it walking you through an imagined line of thinking?

You can prompt an LLM model to provide reasoning first and an answer second and it becomes one and the same.

Worth keeping in mind that all of these points are orthogonal to the quality of reasoning, the bias, or the intentions of the system builders. And building something that emulates humans convincingly, you can expect it to emulate both the good and bad qualities naturally.


In split brain experiments (where the corpus callosum is cut), sometimes the half which is nonverbal is prompted and the action is taken. Yet when the experimenter asks for the explanation the verbal half supplies an (incorrect) explanation. How much of human reasoning when prompted occurs before the prompt? it's a question you have to ask as well.


why does performance improve after chain of thought prompting?

Because a human is measuring it unfairly.

The output without CoT is valid. It is syntactically valid. The observer is unhappy with the semantic validity, because the observer has seen syntactic validity and assumed that semantic validity is a given.

Like it would if the model was alive.

This is observer error, not model error.


Which is a shame. Because car travel in the US is still wildly more expensive than high-speed rail in other countries, while being a vastly inferior form of transportation.


Will this software come with perpetual maintenance updates? For how long? Do I get any new features for free? If not, is it really pay "once"?

Software is never "done". Developers need to be actively involved throughout the lifecycle of a software product. Which, to me, makes this sort of model unsustainable. I'd rather know that software I rely on won't be abandoned once the base of new customers drys up. And I don't mind paying for that.


When buying software like this your perspective should really be "WYSIWYG". You should assume support ends completely tomorrow, and any patches you do get are a windfall.

That means key considerations for every purchase are:

* How easy would it be to migrate from it to something else? * How easy would it be for someone else to develop compatible extensions and drop in replacements? * How easy would it be to troubleshoot and patch it yourself?

Perfectly bug free software is of course difficult. But reasonable level of polish is very doable. It used to be the norm for decades, and they didn't even pay devs as much as now. Modern subscription software does not have fewer bugs than traditional waterfall software.


> Easier refactoring (no wondering who all the consumers of GET /users are)

Unless your API calls are properly abstracted into function calls. No reason you can't just have a `getUsers()` function on the frontend, and then use your favorite “find all references” tool from there.


I got the opposite result. ChatGPT-3.5 won 5-0 for me. For me, LLaMa 2 gave longer answers that sometimes strayed away from the original question.

They both gave great answers overall though.


Same for me. Interesting..


Got a source for that claim?


Common sense?

Government (taxpayers) underwrite unlimited amounts to uninformed buyers paying informed sellers.

I am a seller (college boss), and I can boost my own income, and I know I have no competitive pressure to keep prices down, so why not increase the college budget and spending, hence justifying tuition increases.


Common sense is often wrong. I haven't found any reliable source yet that even mentions the availability government loans as being a driving factor. Based on the limited research I've done over the last few minutes, the main drivers appear to be:

1. A large increase in demand for College. I could imagine this may be, in part, due to more accessible student loans. But it appears that many factors have come into play, and the main one is the perception that you cannot get a liveable wage in this country without a college degree.

2. Less funding from state governments

3. Increased spending in administration and student services unrelated to education.

Here are the sources I've read through so far. https://www.bankrate.com/loans/student-loans/why-is-college-... https://manhattan.institute/article/a-new-approach-for-curbi...


> Based on the limited research I've done over the last few minutes, the main drivers appear to be:

All your drivers cannot cause a price increase if an 18 to 25 year old cannot get unlimited funding from a lender to make a stupid decision.


3 is a side effect, not a primary effect.

In 1 you're talking about a self-reinforcing effect. "I have to put on makeup to look pretty because everyone puts on makeup to look pretty". When everyone believes it, it becomes true.


Common sense is horseshit. You know what was common sense in the 1600s? A gentleman is clean, so he doesn't need to wash his hands between cutting open a corpse and delivering a baby, and oh gee why do so many women get awful infections after childbirth?

Common sense is the rallying cry of people who cannot backup their claim by other means.


Because it is not 'free'. It puts students in debt.

You are boosting your income by exploiting students through usury.

Lets say there is no government, then where was the 'competitive' pressure? How does government loans take away competition? Did they close schools to eliminate the competition?


Good faith reading is that he means its risk free to the school, because they get paid up front even if the student doesn't graduate, much less earn enough to pay back his loan.


Googling it provides a quick-answer box that states this claim and provides a source:

https://www.google.com/search?q=what+percent+of+student+loan...

92% Federal student loans make up the vast majority of American education debt—about 92% of all outstanding student loans is federal debt. The federal student loan portfolio currently totals more than $1.6 trillion, owed by about 43 million borrowers.

I think it's then pretty fair game for a forum comment to then suggest there's a causal relationship between "an entity that can print the currency that denominates the debt is backing the debt" and "this drives the price of the thing up due to easier access to funding for it".


No, it doesn't. That quick-answer doesn't even mention rising tuition costs. It's only tangentially related. The existence of government backed student loans does not in itself prove they are the main cause of increased tuition.

To prove that claim, you'd need to actually demonstrate some sort of causal relationship and have data to back it up

For example, you could make the argument that:

1. Demand is the primary cause of tuition increases

2. Government backed student loans are the primary cause for the increase in demand

I believe that 1 is almost certainly the main cause of tuition cost increases. What I haven't seen any evidence for is 2. If it's true, show me the evidence. I'm open to it.

From what this article states and the research I've done, the main driver of demand is not cheap government loans. It's the labor market. The best jobs in this country require a college degree. That is what is probably driving demand, the desire to earn a liveable wage. Not the fact that you can get a cheap loan.


The article this post was about?


The article does not support this claim.

The claim: > Main cause: Government-backed student loans.

I've read the article. It discusses student loans and how they've skyrocketed, but never makes the claim that their availability is the cause. Instead, it argues a change in America's "labor profile" is to blame for the rise in demand (which, in turn, leads to higher tuition):

> Rising enrollment is also associated with a changing US labor profile; for example, manufacturing jobs were eclipsed by “business and professional services” jobs as well as healthcare, education, and retail jobs.

Going further, the author suggests we need to forgive/cancel the student loans:

> These loans should be forgiven/cancelled

To me, this implies the author does not believe federal student loans are to blame.


Probably not. According to OpenAI, most of its data comes from before 2021:

> GPT-4 generally lacks knowledge of events that have occurred after the vast majority of its data cuts off (September 2021)

https://openai.com/research/gpt-4


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: