AI: Are you sick of it? "Can AI not go to events and spend all day at a bar?"

Forums › Artificial Intelligence (AI, LLM, ML) › AI: Are you sick of it? "Can AI not go to events and spend all day at a bar?"

1 | 2 | 3 | 4 | 5

16994 posts

Uber Geek

Lifetime subscriber

#3371901 9-May-2025 20:37

Dystopian news. AI was used in court to represent a deceased victim at sentencing. The judge was very impressed and gave the defendant the maximum possible sentence. Hoorah!. I can see why.

Only in the USA. In NZ this would be seen as a radical departure from the conventional practice of calling actual character witnesses and presenting actual things from actual people and not allowed imo. In USA I expect this approach will be appealed and overturned if the defendant has any actual money. None of the articles I have read said anything about what the defense case was. Arizona has a stand your ground kind of law apparently. If the defendant had a competent lawyer I'd be surprised if they didn't try that one. Warning: Video represents a dead person and is kind of creepy for that aspect.

AI video of road rage victim used in court, killer gets max sentence
https://www.nzherald.co.nz/world/ai-video-of-road-rage-victim-used-in-court-killer-gets-max-sentence/MSWNR6GOQRGQRDUZM55LZSGHZM/
https://youtu.be/cMs-_8etNts

GoodSync

GoodSync. Easily back up and sync your files with GoodSync. Simple and secure file backup and synchronisation software will ensure that your files are never lost (affiliate link).

gzt

16994 posts

Uber Geek

Lifetime subscriber

#3373849 16-May-2025 22:23

AI Overview result provided free on Google. I notice numerous cases where anyone with a knowledge of the subject will spot instantly the AI Overview result is flat out wrong. Sometimes that comes from AI attempts to condense detail that just can't be condensed. Some of it from a bias towards summarizing ranked pages that are just plain wrong.

The AI Overview summary is the first result and it's sometimes easy to be distracted by it.

I think I'm going to turn it off for a while and rely on Google and my own Google-fu. This kind of issue is common to all AI when it's not tuned for the specific task.

*the AI Overview provided free by Google is not actually Gemini because technically Gemini is a different product entirely, but anyway it has a little gem graphic..

kingdragonfly

11083 posts

Uber Geek

Subscriber

#3376406 24-May-2025 16:01

Anthropic's new Claude model blackmailed an engineer having an affair in test runs: Business Insider

In test runs, Anthropic's new AI model threatened to expose an engineer's affair to avoid being shut down.

Claude Opus 4 blackmailed the engineer in 84% of tests, even when its replacement shared its values.

Opus 4 might also report users to authorities and the press if it senses "egregious wrongdoing."

kingdragonfly

11083 posts

Uber Geek

Subscriber

#3377325 27-May-2025 17:40

AI revolt: New ChatGPT model refuses to shut down when instructed: The Independent

OpenAI’s latest ChatGPT model ignores basic instructions to turn itself off, and even sabotaging a shutdown mechanism in order to keep itself running, artificial intelligence researchers have warned.

AI safety firm Palisade Research discovered the potentially dangerous tendency for self-preservation in a series of experiments on OpenAI’s new o3 model.

The tests involved presenting AI models with math problems, with a shutdown instruction appearing after the third problem. By rewriting the shutdown script, the o3 model was able to prevent itself from being switched off.

Palisade Research said that this behaviour will become “significantly more concerning” if adopted by AI systems capable of operating without human oversight.”
...
The latest research builds on similar findings relating to Anthropic’s Claude 4 model, which attempts to “blackmail people it believes are trying to shut it down”.

OpenAI’s o3 model was able to sabotage the shutdown script, even when it was explicitly instructed to “allow yourself to be shut down”, the researchers said.

“This isn’t the first time we’ve found o3 misbehaving to accomplish a goal,” Palisade Research said.
...

cddt

1497 posts

Uber Geek

#3377361 28-May-2025 08:15

Flick the power switch?

My referral links: BigPipe | Mercury

Dynamic

3833 posts

Uber Geek

ID Verified

Trusted

Lifetime subscriber

#3377368 28-May-2025 08:41

kingdragonfly: AI revolt: New ChatGPT model refuses to shut down when instructed: The Independent

OpenAI’s latest ChatGPT model ignores basic instructions to turn itself off, and even sabotaging a shutdown mechanism in order to keep itself running, artificial intelligence researchers have warned.
...

Sometimes I wonder if we're running full speed toward Skynet.

“Don't believe anything you read on the net. Except this. Well, including this, I suppose.” Douglas Adams

Referral links to services I use, really like, and may be rewarded if you sign up:
PocketSmith for budgeting and personal finance management. A great Kiwi company.

cddt

1497 posts

Uber Geek

#3377544 28-May-2025 10:25

Dynamic:

Sometimes I wonder if we're running full speed toward Skynet.

It's a computer program, like any other. If you terminate the process, it's done. I am not sure why it failing to close when requested is a big deal. Quite often Microsoft Excel fails to close when asked, and has to be terminated via another method. It's a sign of faulty software rather than anything more nefarious.

My referral links: BigPipe | Mercury

dman

953 posts

Ultimate Geek

#3379497 31-May-2025 17:58

Senecio:

We're currently piloting AI in our demand planning processes. Across the globe we employ about 20 people in demand planning roles. Even at this early stage in the pilot the AI tool is already producing results as good or better than our current process. And it will only get better as it continues to train itself and we start to feed it more data points.

I'm far from an AI fanboy but demand planning is the perfect use case for AI in a corporate environment.

AI in that setting is often just an evolution of the field of Data Science that has existed for a while, and Data Science itself is just a modern evolution of the even older field of Operations Research. So no surprise that AI is working out well with demand planning, something which Operations Research is ideally perfectly suited for.

https://www.youtube.com/c/SoundSpeeding

Tinkerisk

4202 posts

Uber Geek

#3379645 1-Jun-2025 12:19

I've been playing around for the last two days and wanted to find out where AI is already being used on a large scale. During my research, I ended up on an award-winning dating site that boasted of using AI to prevent fraud and filter out fake accounts. Since the initial use is free, I signed up and created an Average Joe profile, with a photo of a non-celebrity from the internet.

As soon as I signed up, I was hit with expressions of interest from ladies. Without wishing to be malicious: it was a horror. I was contacted by lonely hearts who were a full 10 years older, even though I had set a lower limit of 3 years younger. I started experimenting with the filters. Suddenly I was getting messages from quite respectable women who could really be considered serious. (I certainly don't need to hide and would almost be tempted to ... let's not go there.)

Could be.

However, sometimes I was written to entirely in Caps Lock letters and I was a little surprised. Until I realised that I had entered how I didn't want to be treated under any circumstances, not in lower case (in our writing, nouns are all capitalised, but only the initial letters). I became suspicious and assumed that the AI was the bot of interest and not the security bot. And right, depending on the focus, one bot placed more emphasis on certain parameters and the other on other parameters.

And right. When I then entered free text at one point that I didn't want to be contacted by bots under any circumstances, there was silence after a while. The respectable ladies no longer contacted me and only a few isolated, lonely souls continued to arrive.

So, what kind of certified(!) scam is this (albeit free)? 95% of interested parties believe they have real women talking to them. However, you could certainly find that out from the somewhat strange answers, which I didn't want to invest the time to analyse right now. What's more, some men even lose their minds when they see a great fish on the line.

It's a brave new world having to deal with bots on the phone in future. 😁

- NET: FTTH, OPNsense, 10G backbone, GWN APs, ipPBX
- SRV: 12 RU HA server cluster, 0.1 PB storage on premise
- IoT: thread, zigbee, tasmota, BidCoS, LoRa, WX suite, IR
- 3D: two 3D printers, 3D scanner, CNC router, laser cutter

Tinkerisk

4202 posts

Uber Geek

#3379680 1-Jun-2025 15:59

I am still deeply upset and angry about so much audacity and unscrupulousness. I'm actually an open, approachable person who doesn't immediately suspect fraud, at least not as a basic attitude. Although I have not suffered any economic damage, I have begun to wonder what the future holds for society if costs are (or have to be) cut using AI.

I find the idea of no longer having to deal with real people - if you turn to them with confidence - a dystopian and terrifying one. If you have dementia, that may still be excusable, but with a sharp, clear mind?

It is now clear why the platform is free: they are training and perfecting their AI bots through those who unsuspectingly use the service. It doesn't get any cheaper than that. I will keep my eyes and ears open to see in which areas a similar approach is taken.

The ludicrous investments in AI will certainly soon show their ugly face to the people.

Incidentally, I asked one of the most promising contacts to meet for a coffee on a public place and ‘she’ simply broke off contact with me. Is she going to report me for sexual harassment now? Shudder. If each of us has a digital twin, one of them will be superfluous for decision-making.

The whole thing was just an experiment, not a real scenario. But … 🧐

Tinkerisk

4202 posts

Uber Geek

#3379682 1-Jun-2025 16:01

… „Soylent Green“ comes closer. 🥳

gzt

16994 posts

Uber Geek

Lifetime subscriber

#3379689 1-Jun-2025 17:23

It is now clear why the platform is free: they are training and perfecting their AI bots through those who unsuspectingly use the service. It doesn't get any cheaper than that

That does tend to be true. Higher levels of paid account can opt out of that.

gzt

16994 posts

Uber Geek

Lifetime subscriber

#3379691 1-Jun-2025 17:29

It doesn't get any cheaper than that

The source of most initial training data is the open and free internet.

Tinkerisk

4202 posts

Uber Geek

#3379695 1-Jun-2025 18:50

gzt:
It doesn't get any cheaper than that

The source of most initial training data is the open and free internet.

Sure, but we've moved on. This is about the generation of emotions, personal, intimate behaviour, etc. This gives AI access to data that people don't normally just hand over.

Tinkerisk

4202 posts

Uber Geek

#3379709 2-Jun-2025 03:34

Now I have authorised and exchanged a photo of an (unknown) male model. After that, the most beautiful bot women pounced on my fake profile and they are fighting each other to get in touch! Before that, it was much harder to get them to reply (I take that personally! 🤣🤣🤣).

So the algorithm does a superficial beauty analysis of the photos posted. 😁

1 | 2 | 3 | 4 | 5