RE: chatGPT for troubleshooting Oracle errors

I believe this is a reasonable link as a starting point to back track on the various cases and results. Of course Rimini Street and Oracle have each their own press releases.  

I don’t have a dog in this fight.  

Tim’s probably correct that none of the breaches of contract rise to the level of a criminal offense. They apparently did not break into Oracle to obtain the files, since they apparently did so on behalf of licensed customers who apparently had every right to hire a third party to obtain and apply the patches for them. My understanding of the rulings is that those lawfully obtained files where then put to unlicensed use. That’s just my layman’s understanding. But it’s pretty easy to read the rulings yourself.  

https://www.cio.com/article/303439/oracle-scores-another-legal-victory-against-rimini-street.html  

As for chatGPT, sorry, I have no content to add.  

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of John Thomas Sent: Friday, January 20, 2023 2:53 PM
To: tim.evdbt_at_gmail.com
Cc: Sandra Becker; Neil Chandler; oracle-l Subject: Re: chatGPT for troubleshooting Oracle errors  

Yes, you're almost certainly correct, at least in terms of US law. I think the Tomorrow Now case was about breach of contract or copyright. I'm no lawyer.  

I am pretty sure Rimini Street cannot supply patches though, because I had vigorous discussions with them about it at an Oracle conference exhibition and I'm sure they were clear they did not do that.

Regards,  

John Thomas  

N.B. Nothing I have said should be construed as legal advice, you need to seek qualified opinion if in doubt about legal issues.      

On Fri, 20 Jan 2023 at 19:28, Tim Gorman <tim.evdbt_at_gmail.com> wrote:

>> Rimini Street can't supply patches without breaking the law, so you're stuck on the version you have.

Not sure if that is true, because then I don't understand how Rimini Street could have even existed for the past 2 decades or so?

But pretty certain that "breaking the law" is the wrong description, and "violating a contract" is a better description of Rimini's risk. It's a civil issue, not a criminal issue; an important distinction. :)

On 1/20/2023 9:49 AM, John Thomas wrote:

Wow.  

Rimini Street can't supply patches without breaking the law, so you're stuck on the version you have. But I guess you understand that. Your boss...?  

I'd keep an eye on the jobs lists.  

Good luck.  

Regards, John  

On Fri, 20 Jan 2023, 17:19 Sandra Becker, <sbecker6925_at_gmail.com> wrote:

To save money, the company canceled our support contract with Oracle a couple of years ago. This means a lot of searching for answers and relying on the kindness of people on oracle-l. They finally contracted with Rimini Street last year, but so far we've been able to find answers before they came back to us with solutions.  

Suffice to say my boss is NOT a DBA. He frequently tries to tell us how to do things when he has no clue what he's talking about. I did try using chatGPT for a couple of the errors we are encountering building our standbys, but they were not very helpful. I still have access to MOS, just can't open tickets, and that has been more helpful, as well as posting on oracle-l.  

Sandy  

On Fri, Jan 20, 2023 at 12:59 AM John Thomas <jt2354_at_gmail.com> wrote:

Hi Neil. I hear what you are saying, but it comes back with some surprisingly well-written and accurate answers to specific questions about Oracle. I wonder if it's read the manuals... and what the legal implications of that are.  

Recommend you sign up for an account and give it a try :-)  

(Not that you'd rely on it for serious answers to production questions, I still suspect the OP's boss is a bit too much of a sci-fi fan, but for curiosity. I think it's doing something other than returning the "average" answer.)

Regards,  

John Thomas    

On Thu, 19 Jan 2023 at 23:24, Neil Chandler <neil_chandler_at_hotmail.com> wrote:

AI’s tend to work on average opinion. If you are happy with your first Google hit giving you the correct answer, then ChatGPT will also be fine. They work on a similar method of answers to google, where popularity and “average” is more important than accuracy.

The more accurate your input, the better your outcomes will be with an AI - it does have some value - but from my attempts to get great answers from ChatGPT I was largely disappointed at how anodyne the answers were. We aren’t there yet.

Neil.
sent from my phone

> On 19 Jan 2023, at 21:26, Sandra Becker <sbecker6925_at_gmail.com> wrote:
>
> 
> My boss is pushing my DBA team to use chatGPT for troubleshooting Oracle errors instead of using google or MOS. Has anyone had experience using it for this purpose? What has your experience been? Would you recommend it as another tool in the toolbox or as a replacement?
>
> Would appreciate any feedback.
>
> --
> Sandy B.
>

--
http://www.freelists.org/webpage/oracle-l





-- 

Sandy B.

 





--
http://www.freelists.org/webpage/oracle-l