• Security incident: ISF was recently accessed by intruders. Please change your password, and change it anywhere else you used it. Read more

Merged Artificial Intelligence

HSBC is still primarily COBOL in its backend mainframe z/OS systems. Java on servers etc.
Many years ago IBM announced Unix on MVS, and our developers were delighted. Mostly because it was possible to recruit youngsters who thought mainframes were from the stone ages.
 
OS/390 Open Edition? First OS to be fully POSIX compliant I believe.
The legacy systems that maintain account balances etc etc are all still in COBOL and updated to adhere to changes in law, Basel2 etc etc. All the newer services pull data from the mainframe using MQ in the main. And written in Java.
Even the COBOL is largely written on a PC using errr was it RTC ? IBMs acquired IDE toolset. And the internal education at least in the UK was pitiful.
 
I always considered it the Ancient Magic. I was surprised to learn that Z-series mainframes still use it.
A lot of work is still done by batch written by our forefathers and their forefathers unto the 10th generation. Usually managed by the batch scheduler OPC as was now TWS.
Even web servers on Z series still run as JCL as that’s what the initiators and JES work on
.
 
All of my supercomputing is done by batch. We just use modern tools and operating systems. I understand not wanting to throw out code that has been painfully debugged over years. Our structural and fluid dynamics systems were originally coded in the 1980s in Fortran and that's what we still use. But I'm surprised that the methods of running those programs is largely unchanged.
 
Dean of Science at Stellenbosch University writes that although AI generates almost all the words for his articles (including this one):

"AI did almost all of the writing. I remain the expert and the thought leader behind the piece."

 
I have been following AI generated tseampunk videos for a year or so. Continuing small incremental improvements edging closer to reality.

Integration with speech is not here as yet but check out this for an example of what can be done :
Hollywood lookout.
 
Dean of Science at Stellenbosch University writes that although AI generates almost all the words for his articles (including this one):

"AI did almost all of the writing. I remain the expert and the thought leader behind the piece."

Presumably he asked the AI whether or not he is a fraud, and was therapeutically reassured by its sycophancy.
 
On the one hand I can understand what he means and not all of us are good writers, so is it wrong to use a tool that can produce more fluent English than he can manage alone? Would we say that using a spell checker and grammar checker is wrong? On the other hand no matter even if he works hard to ensure that there are no hallucinations and the science is correct is the resulting article better because it is more fluent even though it contains none of his own text?

My view is he has gone too far to claim authorship of the articles, he hasn't used a tool to only improve his fluency he's handed over the communication of his ideas to "someone" else. It IS ike having a ghost writer and then claiming you wrote the article or book. You haven't, you may have supplied the basic information and ideas but you aren't the author.

And even in regards to fluency it is only a type of psuedo-fluency, the structure, the vocabulary, the rule of three ;) all produce the most bland and insipid prose you can imagine, it's corporate writing took to its zenith. It has no character, no idiosyncrasies nothing that makes a text interesting and keeps you engaged so why should we even bother to read it in the first place? All it deserves is to be fed into an AI to summarise it!
 
It does have some funny side-effects.

I received a beautifully-written application letter today.

Sadly, the person who sent it to me forgot to remove their request of whatever AI they were using.

The letter started with: "Write a cover letter for X job based on my cv and the position description."

Saved me reading any further.
 
I have been following AI generated tseampunk videos for a year or so. Continuing small incremental improvements edging closer to reality.

Integration with speech is not here as yet but check out this for an example of what can be done :
Hollywood lookout.
View attachment gord.mp4

I took a single frame from around 10:58 in your linked video and then animated it. This 5 second 0.4 megapixel clip took about 45 minutes to generate on my used workstation. It would take around 7.5 hours at FHD (1080p) 2.3 megapixel resolution.

I used the Lightricks LTX-2.3 video model.

I did write the AI character's line (in response to Gord_in_Toronto's post) and gave a little bit of prompting scene desciption and direction! ...and this was the first and, at least as of yet, only take! The smudges on her face were in the original frame.
 

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom