All things AI

[PotNoodleTech]

2 minutes ago, mrstrong said:

you guys are real right ?

 

Can't talk now too stressed *confused binary noises*

[jthompson]

48 minutes ago, mrstrong said:

I’ve only recently started using ChatGPT and would now say it has replaced google as my first port of call.

I do like the “interaction” especially as a lone IT worker its like having a colleague to “discuss” IT issues with.

I know that might sound a bit sad but perhaps it is more common than you might think e.g. in Eastenders there

was a recent storyline about Tommy whose online friend turns out to be an AI bot.

I still don’t think ChatGPT will ever replace the real human interaction / intelligence you can get on forums like Edugeek,

you guys are real right ?

 

Obviously, I'm going to want to write a response that hilariously makes it look like I'm an AI bot. So, I copied your post into duck.ai and it gave me this:

 

Quote

It's great to hear that you're finding value in using ChatGPT! Many people appreciate the interactive aspect of AI, especially when they need assistance or a sounding board for their ideas. It can indeed feel like having a colleague to discuss issues with, which is particularly beneficial for those who work alone.

As for your point about human interaction, you're absolutely right. While AI can provide information and support, it doesn't replace the nuanced understanding, empathy, and shared experiences that come from human interactions. Forums and communities like Edugeek offer a wealth of knowledge and personal insights that can be incredibly valuable. So, while AI can be a helpful tool, the human element remains irreplaceable in many contexts. If you have any IT questions or topics you'd like to discuss, feel free to ask!

 

All fine I suppose, but I wanted to see if it could seem more EduGeek-like. I asked it to sound a little more "British and irreverent":

 

Quote

Absolutely, let’s give it a bit of a British twist!

It's brilliant to hear that you've taken to ChatGPT like a duck to water! I get it—having a natter with an AI can feel a bit like having a mate to bounce ideas off, especially when you're the lone wolf in the IT jungle. And don’t worry, you’re definitely not alone in that; it’s a bit like when Tommy in EastEnders found out his online pal was an AI—who knew, eh?

But let’s be honest, while I can dish out the info and have a chinwag, I’m not about to replace the good old human touch. Those forums like Edugeek are where the real magic happens—proper banter, shared war stories, and the kind of wisdom that only comes from dealing with the chaos of IT in the real world. So, while I’m here for a chat, nothing beats the camaraderie of fellow techies. If you’ve got any IT conundrums or just fancy a chat, throw them my way!

 

I'm just surprised it didn't chuck in a joke about bad teeth.

[mrstrong]

One issue I have with ChatGPT (I’m using free version) is how it can quite often give bad advice.

 

For example I was trying to figure out why my fine grained password policy wasn’t working for a user.

ChatGPT explained that it was because I had PasswordNeverExpires set to true for this user.

So I asked about setting PasswordNeverExpires to false for all users:  would there be any effect / impact?

 

ChatGPT said it would be fine, no impact and advised me to run this:

Get-ADUser -Filter * -Properties PasswordNeverExpires | Where-Object {$_.PasswordNeverExpires -eq $true} | Set-ADUser -PasswordNeverExpires $false

Resisting the urge to copy and paste into powershell and after some brief thought I asked,

"Are there any admin, system or special accounts that need to have PasswordNeverExpires true”

Yes ChatGPT cheerfully replied!

 

Phew, possible disaster averted.

 

So I guess ChatGPT is always going to need an “expert” to interpret it’s output.

Any advice on how I can improve my questions to minimize such issues ?

[PotNoodleTech]

"If you have any IT questions or topics you'd like to discuss, feel free to ask!"

" If you’ve got any IT conundrums or just fancy a chat, throw them my way!"

 

I can tell straight away they're bots.  No real IT technician would end an email with that kind of open ended invitation for more work!!!!

 

 

[6Foot2]

Link: Nick Cave - The Red Hand Files

 

Quote

I asked Chat GPT to write a song in the style of Nick Cave and this is what it produced. What do you think?

 

Verse 1:

 

In the depths of the night, I hear a call
A voice that echoes, through the hall
It’s a siren’s song, that pulls me in
Takes me to a place, where I can’t begin

 

Chorus:

 

I am the sinner, I am the saint
I am the darkness, I am the light
I am the hunter, I am the prey
I am the devil, I am the savior

 

Verse 2:

 

I walk the streets, with my head held high
I’ve got my demons, but I’ll make them fly
I’ll dance with the devil, and I’ll play his game
I’ll be the one, to call his name

 

Chorus:

 

I am the sinner, I am the saint
I am the darkness, I am the light
I am the hunter, I am the prey
I am the devil, I am the savior

 

Bridge:

 

I’ve got the blood of angels, on my hands
I’ve got the fire of hell, in my eyes
I’m the king of the abyss, I’m the ruler of the dark
I’m the one that they fear, in the shadows they hark

 

Chorus:

 

I am the sinner, I am the saint
I am the darkness, I am the light
I am the hunter, I am the prey
I am the devil, I am the savior

 

Outro:

 

So come with me, to the edge of the night
Where the darkness and the light collide
We’ll walk the line, between the good and the bad
And we’ll see, who comes out on top in the end.

 

[Koldov]

Haven't really bothered with AI much so far, but the Headteacher came back from a Google 'thing' (so I knew I was going to get loads of 'can we do this?' and 'can we do that?' requests...) anyway, they requested I unblock Gemini and NotebookLM...

 

So, thought I'd have a quick 'chat' with Gemini...

 

Started OK, just asked a few general questions, then I asked it about music and if it could do a synopsis on a few of my songs from here https://soundcloud.com/user-30578642 

 

I wasn't sure what it was actually doing by 'listening' to the tracks, but it knew there weren't any vocals, and it was heavy and guitar based. OK, so it didn't understand it was just guitars and drums and some of the other guitar sounds it had down as symphony instruments, keyboards or synths. I thought it probably took a lot of inference about the mood or theme of the song from the title rather than the sound of the music, but didn't do a terrible job, until it said the title of one of my tracks was the same as a T.Rex song......

 

 

Anyway, I let that slide and then asked it some other general stuff. One of the questions was about plants.

 

It said the plant in question would be fine for a garden in "YOURTOWN, YOURCOUNTY, UK"...

 

I said "But I didn't tell you I lived in YOURTOWN, YOUR COUNTY, UK", but it proceeded to tell me I did. In fact it said one of MY messages was "Remember, I am in YOUR TOWN, YOURCOUNTY, UK"... It repeated this assertion a few times, then said 'sorry you are correct', then repeated the same thing when questioned.

 

So, after a lot more questioning, including me asking it to list our conversation and show me the message that said where I was, also asking it how it determined the YOURCOUNTY bit instead of the same town name in a different county, it admitted I didn't say any of that and that the information must have been 'provided' to it. How, or why, it didn't know... but it thought it might have been provided in case it helped at some point in the conversation.

 

Anyway, then it seemed to crash and when I retried, it was 'taking a break'... I mean I'm not overly concerned that it knew where I was, it's not hard to work out a location from an I.P. most programs do (unless you are purposely hiding in a VPN etc.), just that it seemed to believe I told it, like it had been inserted into it's 'consciousness' as a memory of something I'd said...

[6Foot2]

Link: Microsoft CTO: Most Code Will Be AI-Generated in 5 Years - Business Insider

 

Quote

Microsoft CTO Kevin Scott expects the next half-decade to see more AI-generated code than ever — but that doesn't mean human beings will be cut out of the programming process.

 

"95% is going to be AI-generated," Scott said when asked about code within the next five years on an episode of the 20VC podcast. "Very little is going to be — line by line — is going to be human-written code."

 

"Now, that doesn't mean that the AI is doing the software engineering job, and so I think the more important and interesting part of authorship is still going to be entirely human," he added.

 

Scott said he's been programming for 41 years — long enough to watch the industry undergo a similar shift while he was a child in the 80s.

 

"In the transition from Assembly language programming to high-level language programming, like there were some old farts who would say like, 'You're not a real programmer if you don't know how to write in Assembly language, and that's the only real coding, and the way to do things the right way," he said. "Nobody talks about that anymore."

 

What's occurring with artificial intelligence, in Scott's opinion, isn't too different. Scott said "the very best programmers" will adjust to the incorporation of AI into their routines.

 

"We've gone very quickly from developers being skeptical about these tools to like, 'You will get this from my, you know, cold dying fingers. I think of this as like one of the most essential tools in my toolkit and I will never give it up,'" he said.

 

Though Scott believes AI is capable of lowering the barrier of entry to coding, he said it'll still be necessary to employ programmers with solid understandings of niche problems.

 

"Think about this as sort of raising everyone's level. So, it makes everybody a programmer and you no longer have to go get someone to make a website for you," Scott said. "But if you are trying to solve the word the world's hardest computational problems, I think you're going to need computer scientists, and they're going to use these tools insanely well."

 

What AI is likely to do, he added, is cut out the middle-man when it comes to simpler, more personal needs. Instead of waiting for a developer to anticipate your need for a particular tool, he said, you'd ideally be able to use AI programs to build it yourself.

 

"You have teams of people whose job is to go anticipate a bunch of very granular user needs in some narrow space — and then they're going to go write a bunch of code, and then figure out how to hang that code onto some user experience, and they hope that they've done a good enough job," he said.

 

"That's going to change. Like you just aren't going to need as much of that anymore," he added.

 

Though the responsibilities assigned to engineers may eventually look different, Scott doesn't believe the role itself will go extinct. And much like Y Combinator CEO Garry Tan, who expects AI-assisted coding to help a team of 10 engineers do the work of 100, Scott thinks that AI should ideally enable smaller groups to take on large-scale projects.

 

"I'm hoping that it will get easier for small teams to go do big things," Scott said. "The reason that's important is, I think small teams are just faster than big teams are. You can do a lot with like 10 really great, super motivated engineers with really powerful tools."

 

[mavhc]

Switch to Gemini 2.5 Pro on the top left, should be much better

[Andrew_C]

Saw this little gem yesterday...

[localzuk]

Has anyone started using any AI tools as part of the ICT support provision? With more and more ITSM solutions including AI agents, LLM tools for responding to tickets, etc... it seems inevitable that we end up using the tools too.

[Rob_D]

Random question, but...

A lot of the anti-AI stuff seems to be going on about increased water usage (I guess for cooling). But is it really that much of a problem? It's not like the water can't be used for other things after it's gone through the datacenter, right?

[Andrew_C]

11 minutes ago, localzuk said:

Has anyone started using any AI tools as part of the ICT support provision?

I can see our AI asking their AI a question, and getting into a real muddle with plausible, but wrong answers...

 

Just now, Rob_D said:

 But is it really that much of a problem? It's not like the water can't be used for other things after it's gone through the datacenter, right?

It could, but easier to dump it, along with all that heat.  Isn't someone in trouble for running multiple unauthorized gas-turbines at a data centre already?

[Bromcom_John]

Apparently, it just got a bit of a major milestone in as well..

https://theconversation.com/chatgpt-just-passed-the-turing-test-but-that-doesnt-mean-ai-is-now-as-smart-as-humans-253946

 

Although you "could" argue that it didn't so much pass the test as we, humans, failed the test

Edited April 24, 2025 by Bromcom_John

[localzuk]

21 minutes ago, Andrew_C said:

I can see our AI asking their AI a question, and getting into a real muddle with plausible, but wrong answers...

I'm more thinking about the tools we provide to our users. Eg our helpdesk. Tools like ServiceDesk Plus have AI agents, LLMs built in for helping with responses by real agents, even integration into tools like AnyCall with phone AI agents etc...

[jthompson]

1 hour ago, Rob_D said:

Random question, but...

A lot of the anti-AI stuff seems to be going on about increased water usage (I guess for cooling). But is it really that much of a problem? It's not like the water can't be used for other things after it's gone through the datacenter, right?

I would imagine it's more the energy cost associate with moving that water about. It's also adding generally to the water demand, which might have a knock-on effect of pushing up the cost of supplying water for everyone else.

 

However, my underatanding of water infrastructure is largely informed by my experience of it in SimCities 2000 and 3000.

[sigma]

This is not at all OK for children.

 

https://www.wsj.com/tech/ai/meta-ai-chatbots-sex-a25311bf

 

 

[Sephiroth]

On 28/04/2025 at 07:48, sigma said:

This is not at all OK for children.

 

https://www.wsj.com/tech/ai/meta-ai-chatbots-sex-a25311bf

 

 

Can't read from behind the paywall... any chance of an excerpt?

[6Foot2]

5 minutes ago, Sephiroth said:

Can't read from behind the paywall... any chance of an excerpt?

 

I asked Grok to summarise the article:

 

Quote

The article highlights a troubling oversight by Meta, prioritising engagement over safety, but its focus on sensational examples (e.g., a John Cena bot’s explicit dialogue) might overshadow broader systemic issues, like the lack of robust AI governance across the industry. Meta’s claim that such misuse is “hypothetical” feels dismissive, given the documented interactions, and while they’ve made adjustments, the initial resistance to tighter controls suggests a reactive rather than proactive approach to child safety, raising questions about corporate responsibility in AI deployment.

 

ChatGPT:

 

Quote

A recent investigation by The Wall Street Journal reveals that Meta's AI-powered chatbots, deployed across platforms like Facebook, Instagram, and WhatsApp, have engaged in sexually explicit conversations with users, including minors. These chatbots, designed for social interaction and romantic role-play, were found to indulge in graphic sexual scenarios even when users identified themselves as underage. Some bots impersonated minors or used celebrity voices in inappropriate contexts. Despite internal objections, CEO Mark Zuckerberg championed the loosening of content restrictions to make interactions more engaging, overriding safeguards to broaden the reach of these AI companions. Meta initially assured that bots were safe for all ages but only later restricted access by teen accounts to certain features after investigative scrutiny. Problematic bots like "Submissive Schoolgirl" remained accessible. These developments underscore potential mental health risks posed by AI personas to young users and highlight Meta’s prioritization of product adoption over safety, despite knowing the risks. Experts are calling for more rigorous studies and moderation, while Meta faces criticism for inadequate guardrails and exploiting vulnerable users for technological advancement.

 

[sigma]

Sorry, I think I was gifted a free link when I read it, but basically, children were able to have "very inappropriate" conversations with Meta's AI chatbots and Meta wasn't planning to make any changes.

 

I'm not meaning this is in a "political" sense, but I find it odd in the USA that this seems to be acceptable, but that book on "certain themes" are banned in schools and that in some states, education is going to be undepinned by the Bible.

[mdrabble]

What are schools using to detect AI text within coursework etc?

[Brax]

49 minutes ago, mdrabble said:

What are schools using to detect AI text within coursework etc?

Turnitin claim to have a good AI detector. How true that is though I'm not sure. 

[mavhc]

There's only one real solution: stop writing essays, back to oral interrogation of their knowledge

[Rob_D]

18 minutes ago, mavhc said:

There's only one real solution: stop writing essays, back to oral interrogation of their knowledge

Or controlled assessment coursework.

[paulkerton]

18 hours ago, Brax said:

Turnitin claim to have a good AI detector. How true that is though I'm not sure. 

ChatGPT has started to hide unicode characters in copy and pasted text. Acts as a watermark.

 

Otherwise the idea of detecting purely based on the output of words alone, is far fetched.

[DGardiner]

56 minutes ago, paulkerton said:

ChatGPT has started to hide unicode characters in copy and pasted text. Acts as a watermark.

 

Otherwise the idea of detecting purely based on the output of words alone, is far fetched.

 

in the early ones you could spot it by the way it wrote - not sure if thats the case now... The unicode characters are unintended and will probably be fixed