Re: How has/will AI impacted the RPGLE community?

[VERNON HAMBERG Owner <vhamberg@xxxxxxxxxxxxxxx>]

I agree completely. This is no different from any data that a company feels can give it a competitive advantage. If you are going to use AI, maybe get a private engine set up.

Cheers
Vern


On Mon, 15 Jul, 2024 at 2:43 PM, Scott Klement <sk@xxxxxxxxxxxxxxxx> wrote:
 

To: midrange-l@xxxxxxxxxxxxxxxxxx

I'd be VERY careful about trusting that.

1) You are providing your code on an open source license, which means 
LEGALLY, IBM has the right to distribute it for free to anyone, even if 
they promise not to.

2) IBM will promise not to distribute your code, but only make it 
available for training the AI.  But if someone should happen to ask the 
AI for a routine you provided, the AI might generate almost identical 
code to what you provided -- since after all, it was trained by your code.

So, if you have code that is truly a competitive advantage or trade 
secret, simply don't submit that code.  Submit everything else -- help 
train the AI -- but don't provide anything that'd be a problem if 
someone else got their hands on it.


On 7/15/24 10:44 AM, VERNON HAMBERG Owner via MIDRANGE-L wrote:
> Rob
>
>
> Glad to hear they are doing that. When I heard about this a few weeks or a couple months ago, one of the topics of concern was that of availability to anyone of things added to the "training" materials.
>
> Cheers
> Vern
>
>
> On Mon, 15 Jul, 2024 at 9:03 AM, Rob Berendt <robertowenberendt@xxxxxxxxx<mailto:robertowenberendt@xxxxxxxxx>> wrote:
>   
>
> To: midrange systems technical discussion
>
> Vern,
> If you read the site, or attended the webinar, there is clearly a box you
> can check that says whether only IBM can read the submitted code or can the
> whole community.  This was created for those who were concerned about
> submitting code with a competitive advantage.
>
> On Wed, Jul 10, 2024 at 4:37 PM VERNON HAMBERG Owner via MIDRANGE-L <
> midrange-l@xxxxxxxxxxxxxxxxxx<mailto:midrange-l@xxxxxxxxxxxxxxxxxx><mailto:midrange-l@xxxxxxxxxxxxxxxxxx<mailto:midrange-l@xxxxxxxxxxxxxxxxxx>>> wrote:
>
> > One major issue is ownership. Things like WATSONX apparently say that
> > everything used to "train" has to be publicly visible - I may not have the
> > right wording.
> >
> >
> > I'm hearing of discussions in at least one context, that companies won't
> > put any of their historical data on any public AI site - why give away to
> > competitors what has taken years to accumulate?
> >
> > So maybe there will come to be private "chatgpt" sites - although if the
> > engine used requires openness, there might be a challenge again.
> >
> >
> > Regards
> > Vern