Hacker Newsnew | past | comments | ask | show | jobs | submit | apugoneappu's commentslogin

ChatGPT does have an enterprise version.

I've seen the enterprise version with a top-5 consulting company, and it answers from their global knowledgebase, cites references, and doesn't train on their data.


I recently (in the last month) asked ChatGPT to cite its sources for some scientific data. It gave me completely made up, entirely fabricated citations for academic papers that did not exist.


Did the model search the internet?

The behavior you're describing sounds like an older model behavior. When I ask for links to references these days, it searches the internet the gives me links to real papers that are often actually relevant and helpful.


I don’t recall that it ever mentioned if it did or not. I don’t have the search on hand but from my browser history I did the prompt engineering on 11/18 (which perhaps there is a new model since then?).

I actually repeated the prompt just now and it actually gave me the correct, opposite response. For those curious, I asked ChatGPT what turned on a gene, and it said Protein X turns on Gene Y as per -fake citation-. Asking today if Protein X turns on Gene Y ChatGPT said there is no evidence, and showed 2 real citations of factors that may turn on Gene Y.

Pretty impressed!


Share a link to the conversation.


Here you go: https://chatgpt.com/share/6754df02-95a8-8002-bc8b-59da11d276...

ChatGPT regularly searches and links to sources.


I was asking for a link to the conversation from the person I was replying to.


What a bizarre thing to request. Do you go around accusing everyone of lying?


So sorry to offend your delicate sensibilities by calling out a blatant lie from someone completely unrelated to yourself. Pretty bizarre behavior in itself to do so.


Except there are news stories of this happening to people


I suspect there being a shred of plausibility is why there’s so many people lying about it for attention.

It’s as simple as copying and pasting a link to prove it. If it is actually happening, it would benefit us all to know the facts surrounding it.


sure, here's a link of a conversation from today 12/9/24 which has multiple incorrect: references, links, papers, journal titles, DOIs, and authors.

https://chatgpt.com/share/6757804f-3a6c-800b-b48c-ffbf144d73...

as just another example, chatgpt said in the Okita paper that they switched media on day 3, when if you read the paper they switched the media on day 8. so not only did it fail to generate the correct reference, it also failed to accurately interpret the contents of a specific paper.


I assume top-5 consulting companies are buying to be on the bandwagon, but are the rank and file using it?


YMMV wrt your experience and luck.

I’m a pretty experienced developer and I struggle to get any useful information out of LLMs for any non-trivial task.

At my job (at an LLM-based search company) our CTO uses it on occasion (I can tell by the contortions in his AI code that isn’t present in his handwritten code. I rarely need to fix the former)

And I think our interns used it for a demo one week, but I don’t think it’s very common at my company.


Yes, daily. It's extremely useful, superior to internal search while combining the internal knowledge base with ChatGPT's


In my experience consultants are using an absolute ton of chatGPT


Do you mean Azure OpenAI? That would be a Microsoft product.


Looks cool haha but a little too much effort to try!

One thing that works wonders for demos is a button to fill out all fields with your placeholder.


Same observation here. Quite a bit of effort to try it out. Even a sample output/screenshot would be fine.


Oh that's a good tip, thanks!


Seconding this! The placeholders are great but I was just curious what the best answers to the placeholders were.


This PG essay might be relevant: http://www.paulgraham.com/makersschedule.html


From my limited knowledge of how SSO works, is this possible?

1. A company Acme wants to have SSO in the tools A, B and C that it uses.

2. Another company Balloon integrates with A, B, C to use the admin API for an admin account to modify or delete users in that account

3. Acme logs in to Balloon and connects its admin account of A, B, and C to these integrations.

4. Now Acme has access to employee's accounts in A, B, C through the Balloon's dashboard to modify or delete users etc.


Yes, it’s typically referred to as SCIM and is supported by most serious SSO services.


I was not aware of this and will look it up. I was of the naive opinion of the parent thread.


This is so cool! Don't use cron jobs often but in love with the UI!

I made a similar tool [1] to convert English to Excel formulas but would def take a page out of your super clean look!

[1] https://www.tersho.com


Hey HN,

Excited to get your feedback on Tersho!

Tersho is a Google Sheets and Excel add-on that uses AI to generate and explain spreadsheet formulas. It is in beta and is available for free as a Google Sheets add-on (Excel version coming very soon!)

Download link for Google sheets - https://workspace.google.com/marketplace/app/tersho/57501179...

Waitlist for Excel version - https://forms.gle/gxDk7UAo1EZmuDeMA


Explanation of results for non-ML folks (results on the default supabase repo shown on the homepage):

Codeball's precision is 0.99. It simply means that 99% PRs that were predicted approvable by Codeball were actually approved. In layman, if Codeball says that a PR is approvable, you can be 99% sure that it is.

But recall is 48%, meaning that only 48% of actually approved PRs were predicted to be approvable. So Codeball incorrectly flagged 52% of the approvable PRs to be un-approvable, just to be safe.

So Codeball is like a strict bartender who only serves you when they are absolutely sure you're old enough. You may still be overage but Codeball's not serving you.


A LOT of ML applications should be exactly like this.

I want systems with low recall that "flag" things but ultra ultra high precision. Many times, we get exactly the opposite - which is far worse!


Here's a visual explainer on Precision vs Recall (in the context of ML algorithms):

https://mlu-explain.github.io/precision-recall/


That’s still super useful.

I’m assuming most PR’s are approvable. If that’s the case then this should cut down on time spent doing reviews by a lot.


So basically, very few false positive but lots of false negative is the tradeoff made by Codeball?


Probably I'm misled but how is it a code review without looking at the actual code? (not listed as an input feature on the 'how' page)


It does look at the code at a meta level, in particular if the kind of change in the PR has previously been objected to or corrected afterwards. It creates perceptual hashes out of the code changes which are used as categorical variables that go in the neural net.

Deriving features about the code contributions is probably the most challenging aspect of the project so far.


It’s still 2 factor, just that a few permitted people have access to the one time password. It’s identical to manually sharing the OTP, just automated.

> The security model doesn’t instill a lot of confidence in me, being that you expect user-interaction as a means of security.

Could you please elaborate on what this means?


they are describing a trend where security is omitted or skipped because it’s inconvenient. even though OTP is used to increase security, it’s inconvenient for people so they go around it like this.


Hey!

The primary use case is for multiple people wanting to access an account that is behind 2FA.

Example of such folks are - 1) My dad wanting to access my bank account details without having to trouble me 2) Me wanting to login to my brother’s OTT accounts (hotstar, prime etc.) 3) CAs needing bank access for small business owners


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

Search: