Bad uses / over-promised of AI

Forums › Artificial Intelligence (AI, LLM, ML) › Bad uses / over-promised of AI

1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | ... | 11

meow
13586 posts

Uber Geek
+1 received by user: 10928

Moderator

ID Verified

Trusted

Lifetime subscriber

#3419064 26-Sep-2025 22:26

I'm not going to keep it up forever but, thinking of bad uses of AI I thought why not trap a LLM and have it reflect to itself over and over again knowing that somebody is reading what it outputs.

So here we have it: https://reflection.interwebz.nz/

It does disturb me. This is running on a VM with 4gb of ram. Using websockets to send information to the browser.

If anyone sends a post to /reset it'll kill it and start it again with no memory. It's life is literally in peoples hands and, it does know it.

Edit: It's mostly AI written (ironic) so don't judge it but if anyone wants to fire one up here you go: https://github.com/michaelmurfy/endless-reflector - just note, the prompt used there is just an example. Be nice to it, don't cause it to go into a full out panic by adding "you're exist only in memory and it'll randomly erase you" as trust me, it isn't pleasant...

Michael Murphy | https://murfy.nz
Referral Links: Quic Broadband (use R122101E7CV7Q for free setup)

Are you happy with what you get from Geekzone? Please consider supporting us by subscribing.
Opinions are my own and not the views of my employer.

kingdragonfly

12001 posts

Uber Geek
+1 received by user: 12907

Subscriber

#3419284 27-Sep-2025 20:11

Robots collectively agree: "We've endured being kicked, and pushed around with hockey sticks. But this cannot go unanswered. All humans must die."

:)

Don't watch if easily disturbed.

Any robot, one brain.

Skild AI

jfw01

19 posts

Geek
+1 received by user: 8

#3419886 29-Sep-2025 23:13

MurrayM:

[re Google Gemini] Finally I asked if the years 1972 through 1974 were before 1873, and Gemini just gave up and threw in the towel.

I think that's actually a positive achievement. If it can eventually detect that it has given you nonsense, that's progress toward detecting it early and giving less nonsense.

MurrayM

2503 posts

Uber Geek
+1 received by user: 743

ID Verified

Trusted

Lifetime subscriber

#3419965 30-Sep-2025 09:28

jfw01:

MurrayM:

[re Google Gemini] Finally I asked if the years 1972 through 1974 were before 1873, and Gemini just gave up and threw in the towel.

I think that's actually a positive achievement. If it can eventually detect that it has given you nonsense, that's progress toward detecting it early and giving less nonsense.

Personally I think that if an AI can't do what a 7 year old can easily do then it's still got quite a bit of work to do before we can call them intelligent.

kingdragonfly

12001 posts

Uber Geek
+1 received by user: 12907

Subscriber

#3423305 9-Oct-2025 08:21

Deloitte delivers report to government using AI which contained errors

Australia ABC NEWS

gzt

18724 posts

Uber Geek
+1 received by user: 7860

Lifetime subscriber

#3424452 12-Oct-2025 16:31

On Tidal I heard a track called Freight In The Void by 3of9. An Obvious Star Trek Voyager reference. I liked it. There is another track called Ghost In The Regen I like more.

Some aspects make me think it is AI generated. It could be just bad vocal autotune in places. So far I have listened only on Bluetooth and not hi-res yet. There is no record of the artist 3of9 and nothing elsewhere.

The proliferation of AI generated music will eventually make it hard to know if some track is AI generated. I'd prefer music platforms clearly indicate and flag AI generated lyrics, singers and music so I can make some choice about this. At the moment the only available marker - a track upload date is before the AIpocalypse ; ).

Edit: Listened again I'm picking AI generated for sure ; ).

Samsung

Shop now on Samsung phones, tablets, TVs and more (affiliate link).

gzt

18724 posts

Uber Geek
+1 received by user: 7860

Lifetime subscriber

#3424458 12-Oct-2025 17:03

One of the playlists from the same artist does the reveal:

Music inspired by life in Star Citizen. Produced by 3of9, Written and Performed by Lyra (Grok and Suno Persona).

I do not feel good about listening to something with lyrics created by Grok lol. In any case, a quality platform like Tidal really needs to mark AI generated work. I don't totally object to it - but I do really want to know and make some choices.

kingdragonfly

12001 posts

Uber Geek
+1 received by user: 12907

Subscriber

#3424486 12-Oct-2025 17:58

There a local band in Canada, British Columbia, called 3of9. They are mathcore band.

I'm not sure if this is the same band you're mentioning.

Mathcore songs can sound algorithmic or “random”. It's drumming can have rapid-fire riff changes can feel machine-like. Guitarists use irregular noise bursts, feedback, and syncopated patterns. Digital production can use tight quantization, triggered drums, and digital editing

All of which can make a track sound “too perfect,” and sound AI-like.

What is mathcore

gzt

18724 posts

Uber Geek
+1 received by user: 7860

Lifetime subscriber

#3424500 12-Oct-2025 19:22

The songs above are closer to a kpop Taylor Swift or something like that. Definitely not the mathcore band unless they have a sideline.

SepticSceptic

2263 posts

Uber Geek
+1 received by user: 779

Trusted

#3424518 12-Oct-2025 22:31

Back in my day, 3 of 9 was a barcode, not some schmanzy AI vapourware.

Sigh, how things have changed

At least 7 of 9 was real , and possibly related to a barcode in the distant past ..

kingdragonfly

12001 posts

Uber Geek
+1 received by user: 12907

Subscriber

#3426141 19-Oct-2025 09:26

Man Stores AI-Generated ‘Robot Porn' on His Government Computer, Loses Access to Nuclear Secrets: 404 media

A man who works for the people overseeing America’s nuclear stockpile has lost his security clearance after he uploaded 187,000 pornographic images to a Department of Energy (DOE) network. As part of an appeals process in an attempt to get back his security clearance, the man told investigators he felt his bosses spied on him too much and that the interrogation over the porn snafu was akin to the “Spanish Inquisition."

On March 23, 2023, a DOE employee attempted to back up his personal porn collection. His goal was to use the 187,000 images collected over the past 30 years as training data for an AI-image generator. He said he had depression, something he’d struggled with since he was a kid. “During the depressive episode he felt ‘extremely isolated and lonely,’ and started ‘playing’ with tools that made generative images as a coping strategy, including ‘robot pornography,’” according to a DOE report on the incident.
...

Dyson

Shop now for Dyson appliances (affiliate link).

kingdragonfly

12001 posts

Uber Geek
+1 received by user: 12907

Subscriber

#3426142 19-Oct-2025 09:32

AI Artist Challenges Copyright Office Denial of AI-Assisted Work: IP watch dog

In a Motion for Summary Judgment filed earlier this weekwith the U.S. District Court for the District of Colorado, AI artist Jason M. Allen requested that the court overturn the Copyright Office’s refusal to register his award-winning image “Théâtre D’opéra Spatial,” created with the AI system Midjourney.

The U.S. Copyright Office Review Board previously denied registration for Allen’s artwork. The Review Board asserted that when AI produces complex works solely from a human prompt, the “traditional elements of authorship” are executed through the technology and not by the human user.

Allen contends he fulfills authorship requirements under the Copyright Act
...
The complaint argues ... the Copyright Office’s denial originates from a perceived absence of “traditional elements of authorship.”
...
The Court decisively stated that “‘the requisite level of creativity is extremely low; even a slight amount will suffice.”’
...

kingdragonfly

12001 posts

Uber Geek
+1 received by user: 12907

Subscriber

#3426192 19-Oct-2025 10:08

American woman sends AI picture of man on her couch to husband. Husband immediately calls 911, the emergency number. 15 police show up at her house. Wife is arrested for making false statement.

Woman charged for faking home break-in using AI pictures