* Posts by vmy2197

9 publicly visible posts • joined 1 Oct 2018

Judge crosses out some claims by writers against OpenAI, lets them have another crack at it

vmy2197

the main problem is with whoever wrote the planitiffs complaint.

So the judge was not convinced ChatGPT actually reproduced the complainants works. If that's the case the plaintiffs must not had very good examples. Always take a screen capture folks, that's the lesson here. And get a better lawyer.

California approves lavatory-to-faucet water recycling

vmy2197

Re: Please could you ....

Acre feet is the standard unit of measure for bulk water (in reservoirs and rivers,) in the western U.S.. As you might guess, it's the amount of water needed to cover an acre to one, huh, foot.

One acre = 43,560 square feet or, for the rest of the world, 4046.86 square meters.

One acre foot = 325,851 U.S. liquid gallons or 1233.48 cubic meters.

Olympic swimming pools are 50 meters long, 25 meters wide, and 2 meters deep. In terms of volume, when full, these pools hold 2.5 million liters of water or about 660,000 gallons. (from the interweb). Just shy of 2 acre feet.

IBM's vintage Db2 database jumps on AWS's cloud bandwagon

vmy2197

Re: Db2 for z/OS is what we need on AWS!

Rick Mo? Where have I heard that name before?

But you know what they say, there's no substitute for hardware. Even if the Db2Z software could be ported to run on n x86 chip without an interpreter how do you provide the IO bandwidth, the context switching and RAS functions provided by z hardware? Look at how x86 native databases have tried to compensate for the hardware limitations; by scaling out. But scaling out brings it's own problems, chief among them consistency. Eventual consistency across the replicas? Guaranteed consistency with the associated performance hit? Db2 on z is the success it is because of the hardware and z hardware is all about dense computing. You'll never get the same result if you leave the hardware behind and just move the software.

Now IBM sued for age discrim by its own HR veterans

vmy2197

Re: HR...

It was easier to be a paternal company with jobs-for-life when they had a hardware monopoly. Fortune 500 are still using that hardware and the software that goes with it but every company outside the 500 (and probably most outside the Fortune 100,) are doing their best to get rid of the old iron and nobody is writing new code for it(the old stuff.) IT tech is a much more competitive environment now, brutally so. Software is pretty fungible and the move to cloud tech has been devastating for old-line on-premise software sellers.

Authors Guild sues OpenAI for using Game of Thrones and other novels to train ChatGPT

vmy2197

Re: It doesn't store the original, just 'interesting' features of the original

While not OpenAI this story from February, 2023, illustrates how models can store the data they were trained on and can be coaxed to respond with that data. In this case it's an image AI app but the same thing occurs with OpenAI. Which points to a security problem with these models if they've been trained on sensitive data. In a way this reminds me of the early days of the web where developers were allowing unedited, unbounded user input to be fed to legacy backend systems and mid-range databases. With AI models we have a large opaque blob of code and data with little understanding of how it might behave given the right input.

https://www.theregister.com/2023/02/06/uh_oh_attackers_can_extract/

vmy2197

Re: It doesn't store the original, just 'interesting' features of the original

The complaint makes exactly this point. That not only did OpenAI use unlicensed copyrighted works to train the model but in addition the model stores substantial amounts of the unlicensed copyrighted work which it uses to generate responses.

https://www.courtlistener.com/docket/67810584/authors-guild-v-openai-inc/

88. Until very recently, ChatGPT could be prompted to return quotations of text from

copyrighted books with a good degree of accuracy, suggesting that the underlying LLM must

have ingested these books in their entireties during its “training.”

89. Now, however, ChatGPT generally responds to such prompts with the statement,

“I can’t provide verbatim excerpts from copyrighted texts.” Thus, while ChatGPT previously

provided such excerpts and in principle retains the capacity to do so, it has been restrained from

doing so, if only temporarily, by its programmers.

90. In light of its timing, this apparent revision of ChatGPT’s output rules is likely a

response to the type of activism on behalf of authors exemplified by the Open Letter addressed to

OpenAI and other companies by Plaintiff The Authors Guild, which is discussed further below.

91. Instead of “verbatim excerpts,” ChatGPT now offers to produce a summary of the

copyrighted book,

Amazon says Elon Musk's wicked, wicked ways mean SpaceX's Starlink 2.0 should not be allowed to fly

vmy2197

Re: Space is BIG and satellites aren't

Musk is creating a huge collision problem with Starlink and should be stopped. It's like Teslas on autopilot in space.

https://www.space.com/spacex-starlink-satellite-collision-alerts-on-the-rise

This one weird trick turns your Google Home Hub into a doorstop

vmy2197

Re: Google being rather disingenous

On the same WiFi network? You mean like a malware infected WiFi router?

https://www.tomsguide.com/us/russian-router-malware,news-27288.html

WWII Bombe operator Ruth Bourne: I'd never heard of Enigma until long after the war

vmy2197

Women code breakers in the U.S. during the Cold War

Women in code breaking efforts during the Cold War

https://www.smithsonianmag.com/history/women-code-breakers-unmasked-soviet-spies-180970034/?no-ist