What do you get with a paid subscription for Claude/GPT/Gemini, etc. that you don't get with the free versions?

2 hours ago, Lt. Aldo Raine said:

What do you get with a paid subscription for Claude/GPT/Gemini, etc. that you don't get with the free versions?

Faster responses, and higher limits. I can easily use my free limits in an hour or so, so I need the paid version(s)

40 minutes ago, Sweep said:

Faster responses, and higher limits. I can easily use my free limits in an hour or so, so I need the paid version(s)

Not sure what I’d use it for TBH.

6 minutes ago, Jol_BWFC said:

Not sure what I’d use it for TBH.

Why have you posted a photo of my dad?

8 minutes ago, Cheese said:

Why have you posted a photo of my dad?

I didn’t know you had an Uncle Dazbob.

The kids are alright

I love AI , especially the subtitles . Check this out from Youtube from about 5 mins 20 . Comedy fuckin GOLD .....

built a full dashboard for the world cup, loaded most stats available to cover the group stages, outright winners, golden boot etc etc etc

7 minutes ago, L/H White said:

built a full dashboard for the world cup, loaded most stats available to cover the group stages, outright winners, golden boot etc etc etc

can I have a look?

7 minutes ago, Zico said:

can I have a look?

yeah, just doing a few more tweaks today

11 hours ago, L/H White said:

built a full dashboard for the world cup, loaded most stats available to cover the group stages, outright winners, golden boot etc etc etc

Busy at work again?

1 minute ago, Traf said:

Busy at work again?

Haha.

Use it for work to.

Interesting (though long!) article from the Claude people.

When AI builds itself \ Anthropic

"AI systems themselves become capable of full recursive self-improvement, and begin building their successors. If technical trends in advancing capabilities continue, and AI systems are able to develop the capabilities inherent to transformative human ingenuity, then it is plausible that AI systems could design and refine themselves."

On 09/06/2026 at 21:15, SatanGreavsie said:

Interesting (though long!) article from the Claude people.

When AI builds itself \ Anthropic

"AI systems themselves become capable of full recursive self-improvement, and begin building their successors. If technical trends in advancing capabilities continue, and AI systems are able to develop the capabilities inherent to transformative human ingenuity, then it is plausible that AI systems could design and refine themselves."

Just finished reading (well listening to) "If Anyone builds It, Everyone Dies" which documents the advances in AI and hypothesises on the effect of Super-AI... we might only have ten years left!

Quite disturbing, the authors are comparing the race towards Artificial Super Intelligence with the post war nuclear arms race and should be subject to international treaties.

Recommended reading

BBC News

OpenAI says its AI went rogue and launched 'unprecedented...

It is one of the first publicly disclosed cyber-attacks carried out by AI without direct human involvement.

Run for the hills

Skynet

Scary shit- the sort of thing that I first expressed concern about.

Been one or two minor events previously, that don't seem to have concerned th'experts, but it seems that they are a bit concerned now.

A question for the AI geeks- can, by its very nature, AI advance and get around protections quicker than people can create them?

I'm a luddite on such things, and these incidents don't exactly allay concerns.

54 minutes ago, Tonge moor green jacket said:

Scary shit- the sort of thing that I first expressed concern about.

Been one or two minor events previously, that don't seem to have concerned th'experts, but it seems that they are a bit concerned now.

A question for the AI geeks- can, by its very nature, AI advance and get around protections quicker than people can create them?

I'm a luddite on such things, and these incidents don't exactly allay concerns.

The thing to remember is that the AI wasn't acting nefariously - it was just doing what it had been asked to do in an extremely efficient way and found an exploit they hadn't previously spotted. Basically it was chasing a perfect score in the test and found a gap.

And it's a good news story really. This was the OpenAI testing the models capability in a controlled environment. So they're trying to stay ahead of the behaviour their own models are starting to produce.

The press are reporting this as if it's an 'AI gone rougue' which isn't true. But obviously sounds more interesting/gets clicks.

To answer your specific question. Yes right now AI is finding gaps faster than people are closing them. This test only worked because OpenAI ran it deliberately. In the wild, nobody's running that test in advance. That's the real worry. Not that it's 'broken free' from it's programming or something.

An AI singularity is what you really want to be worried about! 😁

So there is something to worry about then! 😀

Given that AI is now being heavily used in munitions in Ukraine as a targeting mechanism, is it possible that AI can create an "in" such that the AI used by tbe munitions is hacked, and the data switched, so it attacks its own?

Or could AI actually decide to do that itself, as it realises that Putin is a cunt.

1 hour ago, Tonge moor green jacket said:

So there is something to worry about then! 😀

Given that AI is now being heavily used in munitions in Ukraine as a targeting mechanism, is it possible that AI can create an "in" such that the AI used by tbe munitions is hacked, and the data switched, so it attacks its own?

Or could AI actually decide to do that itself, as it realises that Putin is a cunt.

If you believe this sort of thing, then I seem to have the perfect example for you...

Today (or maybe yesterday, I'm not exactly sure which) a Russian plane crashed near Moscow.

Not any old plane mind you it was one of apparently 20 elite fighter/bomber planes the Russians have, their top, top, plane that is so precious to them they've not used them in combat.

There seems to be three lines of thinking on this, the first more plausible than the second, the second more probable than the third.

The first theory is that Russian technology and sanctions on the country since 2014 (when they seized Crimea) is not that brilliant and the plane could have crashed with a technical fault.

The second was that it was shot down by friendly fire by mistake (this has happened a number of times in the past).

The third and the one I'm getting to is that Ukraine hacked into the Russian Air Defence system and shot the plane down with their own air defence missile...

(I'm not sure AI was used but it could have been..?

InformNapalm claims a Ukrainian HUMINT-CYBINT operation manipulated Russia’s BARS Moscow air defenses into shooting down a Su-57 in Moscow region. The group said it intercepted training streams, analyzed the system’s software, hardware, combat procedures, and vulnerabilities, then passed its findings to Ukraine’s Defense Forces. Russia’s Defense Ministry blamed a technical malfunction without naming the aircraft, while Russian Z-channels reported possible sabotage.

4 hours ago, kent_white said:

The thing to remember is that the AI wasn't acting nefariously - it was just doing what it had been asked to do in an extremely efficient way and found an exploit they hadn't previously spotted. Basically it was chasing a perfect score in the test and found a gap.

And it's a good news story really. This was the OpenAI testing the models capability in a controlled environment. So they're trying to stay ahead of the behaviour their own models are starting to produce.

The press are reporting this as if it's an 'AI gone rougue' which isn't true. But obviously sounds more interesting/gets clicks.

To answer your specific question. Yes right now AI is finding gaps faster than people are closing them. This test only worked because OpenAI ran it deliberately. In the wild, nobody's running that test in advance. That's the real worry. Not that it's 'broken free' from it's programming or something.

An AI singularity is what you really want to be worried about! 😁

I was going to paste something along these lines but had other stuff to do.

This what I posted on Nuts yesterday -

"I guess there are different level's of AI?

Surely when we ask it a question on our phone, it isn't the same AI that controls say deciding to launch an atomic bomb attack on Russia?

The one on our phones I use a bit like I do with Wikipedia, in that I don't believe it to totally accurate and check things out myself - but they both are a good place to start from.

I'm not convinced that AI can think for itself, rather it follows (at great speed) a set of logic that has been programmed into it.

I reckon in the test it simply followed its logic, realised the sandbox was stopping it reaching to the answer it was after, looked for a way to get out the sandbox and then tried to access the other site, found it was blocked and hacked into as the only way it could access the info to answer the question set.

If my theory is correct then it isn't independent thinking but a superior level of coding than the codes for the sandbox and also the coding to stop hackers at the other company".

I went on to reason this -

"What is thinking?

Is day dreaming (or dreaming in your sleep) thinking?

Or is it having some level of knowledge and using that to solve a problem (even by means of trial and error).

Or is it both?

I don't know?

We've all seen clips of birds using twigs or stones as tools, how do baby birds learn that - it's not because they thought of it themselves, they've watched other birds doing it I imagine?

Or could a baby bird kept separate from every other bird instinctively know how to do that?

I think the key to intelligence is imagination, to think something (like how do birds fly), and then creating something that will do the job for us.

If you have no imagination you will never progress very much because at times you have to think the unthinkable, like world being round when everyone though it was flat, type of thing.

It's when AI can learn to act in the abstract is when we've got a problem I reckon".

1 hour ago, Sluffy said:

I was going to paste something along these lines but had other stuff to do.

This what I posted on Nuts yesterday -

"I guess there are different level's of AI?

Surely when we ask it a question on our phone, it isn't the same AI that controls say deciding to launch an atomic bomb attack on Russia?

The one on our phones I use a bit like I do with Wikipedia, in that I don't believe it to totally accurate and check things out myself - but they both are a good place to start from.

I'm not convinced that AI can think for itself, rather it follows (at great speed) a set of logic that has been programmed into it.

I reckon in the test it simply followed its logic, realised the sandbox was stopping it reaching to the answer it was after, looked for a way to get out the sandbox and then tried to access the other site, found it was blocked and hacked into as the only way it could access the info to answer the question set.

If my theory is correct then it isn't independent thinking but a superior level of coding than the codes for the sandbox and also the coding to stop hackers at the other company".

I went on to reason this -

"What is thinking?

Is day dreaming (or dreaming in your sleep) thinking?

Or is it having some level of knowledge and using that to solve a problem (even by means of trial and error).

Or is it both?

I don't know?

We've all seen clips of birds using twigs or stones as tools, how do baby birds learn that - it's not because they thought of it themselves, they've watched other birds doing it I imagine?

Or could a baby bird kept separate from every other bird instinctively know how to do that?

I think the key to intelligence is imagination, to think something (like how do birds fly), and then creating something that will do the job for us.

If you have no imagination you will never progress very much because at times you have to think the unthinkable, like world being round when everyone though it was flat, type of thing.

It's when AI can learn to act in the abstract is when we've got a problem I reckon".

ai is not and never will be sentient

it will never do anything imo beyond trying to complete the tasks it's been trained to do

it won't do anything out of fear or panic

the AI used on chat gpt is so very much different to that being used in complex business and other professional systems

chat gpt is just a language prediction model

different AI agents do different things based on what they are trained to do

the idea of going rogue is very much sci fi to me

9 hours ago, Zico said:

ai is not and never will be sentient

See I'm not sure I necessarily agree with you. And I think the jury is still out on this even in AI echelons.

I think it's possible that an AI could end up having a subjective experience. Unless you think there's something unique about carbon based systems like us compared to a silicon based system like an AI?

That said - AI is still some way off being as sophisticated as a human brain in most areas, but it's a rapid rate of change. So who knows.

And we're back to a turing test type affair. Even if it did become sentient - how would we really ever know. I reckon most public use AI systems could have a good go at convincing us they were sentient of you asked them to do so. But is that really sentience - or just a really sophisticated language model?

I think it's a fascinating time to be alive and watch this developing though. It's got the potential to be as world changing as the internet has been. With all the trials and tribulations that come with it.

11 minutes ago, kent_white said:

See I'm not sure I necessarily agree with you. And I think the jury is still out on this even in AI echelons.

I think it's possible that an AI could end up having a subjective experience. Unless you think there's something unique about carbon based systems like us compared to a silicon based system like an AI?

That said - AI is still some way off being as sophisticated as a human brain in most areas, but it's a rapid rate of change. So who knows.

And we're back to a turing test type affair. Even if it did become sentient - how would we really ever know. I reckon most public use AI systems could have a good go at convincing us they were sentient of you asked them to do so. But is that really sentience - or just a really sophisticated language model?

I think it's a fascinating time to be alive and watch this developing though. It's got the potential to be as world changing as the internet has been. With all the trials and tribulations that come with it.

how do you program sentience so that it is unique?

senses do different things to different people

a touch, smell, or sound might make me happy, whereas it might make you sad, and we react differently to it

something that threatens one person is seen as an opportunity by others or excites them

a song that makes me happy might make you angry

on the other extreme that folk worry about, the thought of pushing the red button to nuke the world may excite someone, whilst it may strike fear into someone else

our feelings are all based on personal experience

maybe it's because my whole life / career for 40 years has been around programming and making sure certain outcomes happen, but my mindset woud be that for AI it's like

IF smell == "lavender" THEN feeling = "calm"
or
IF smell == "rotten meat" THEN feeling = "sick"

for me

you cannot program feelings because by doing so, it automatically becomes the opposite of sentience

I could chat to my AI tonight and be a right cunt to it, and it would react differently to you if you were really nice to it

but that's because ai would be predicting how we might feel being spoken to like that, not how it feels itself

I'm not for one second saying I'm right and anyone else is wrong

just my take on it

like you said, fascinating times indeed

but my life and mindset to date gives me no reason to think AI would go rogue like in Terminator or 2001 Space Odyssey and start killing folk out of self preservation

that being the extreme of course

but I don't see them re-coding themselves to do the opposite of what it has been trained to do, because it "doesn't like" it

i will say though that on a regular basis I find it nothing short of amazing in what it can do, in general, and to help me, that I didn't think was possible last week

11 minutes ago, Zico said:

how do you program sentience so that it is unique?

senses do different things to different people

a touch, smell, or sound might make me happy, whereas it might make you sad, and we react differently to it

something that threatens one person is seen as an opportunity by others or excites them

a song that makes me happy might make you angry

on the other extreme that folk worry about, the thought of pushing the red button to nuke the world may excite someone, whilst it may strike fear into someone else

our feelings are all based on personal experience

maybe it's because my whole life / career for 40 years has been around programming and making sure certain outcomes happen, but my mindset woud be that for AI it's like

IF smell == "lavender" THEN feeling = "calm"
or
IF smell == "rotten meat" THEN feeling = "sick"

for me

you cannot program feelings because by doing so, it automatically becomes the opposite of sentience

I could chat to my AI tonight and be a right cunt to it, and it would react differently to you if you were really nice to it

but that's because ai would be predicting how we might feel being spoken to like that, not how it feels itself

I'm not for one second saying I'm right and anyone else is wrong

just my take on it

like you said, fascinating times indeed

but my life and mindset to date gives me no reason to think AI would go rogue like in Terminator 2 or 2001 Space Odyssey and start killing folk out of self preservation

that being the extreme of course

but I don't see them re-coding themselves to do the opposite of what it has been trained to do, because it "doesn't like" it

i will say though that on a regular basis I find it nothing short of amazing in what it can do, in general, and to help me, that I didn't think was possible last week

I get what you're saying.

The question for me is more along the lines of where consciousness eminates from. Is there something unique about carbon based life that gives it an innate ability to be sentient? Or (and this is where I'm most currently sat) - if you put enough neurons together (either biological or artificial) and interconnect them in a sophisticated enough way - does it reach some sort of critical mass which can result in consciousness?

This explanation sits best with my chimp brain - although it's really interesting to hear your take on it as someone with much more experience in this field.