Bring on the AI swarms. They’re the only thing that can defend us now that AI is free

Unpredictable token bills haunted the first half of 2026 as AI experimenters pushed deeper into the tech and often found useful results could only be achieved after surprising and unwelcome expenditure. We've gone from tokenmaxxers to tokenmisers, counting every character we send to AI services. That changed the way we used AI, …

[amanfromMars 1]

For Second Rate Nations being led by Sub Prime Ministerial Post Officers .... The Naked Ape Truth

Bring on the AI swarms. They’re the only thing that can defend us now that AI is free

Frontier-level AI runs everywhere, so attacks will never stop

Have you sent that directly to the government dunces pratting around in the likes of a Westminster and 10 Downing Street, Mark?

Although I might very well agree with you that they would not agree with you and fail to see the evidence of all that surrounds them. Gotta keep the taxing muppet show going and rolling on down that rocky ole road to nowhere worth arriving at and enjoying.

[Pascal Monett]

We're not ready for it

But that's not exactly a problem, since computing advances will wait for people finding a use for them.

It's not like there's a law that says that everyone has to use some new bit of software.

Also, this is El Reg. Please stop endowing this pseudo-AI hype with the moniker AI.

It is not AI. It is a statistical analysis machine, and there aren't many people who need that in their daily lives.

[You aint sin me, roit]

Re: We're not ready for it

You say that, but humans are ingenious. Give them a tool and they will find something to do with it.

No, they don't need it. No, it might not make their lives better. Yes, they will use it, and probably in ways we haven't yet imagined.

Just look at FaceBook, even grannies are addicted to it.

[Dinanziame]

Re: We're not ready for it

It is not AI. It is a statistical analysis machine, and there aren't many people who need that in their daily lives.

I guess everybody needs a hobby

[Pete 2]

Statistics or facts?

> It is not AI. It is a statistical analysis machine

Statistically, someone on this forum was going to make that statement. If you claim intelligence for yourself, then based on your own output, machines which do the same should be given the same description.

[Throg]

Re: Statistics or facts?

Of course. And when "the machines" do that, we will.

[Anonymous Coward]

Re: Statistics or facts?

The fact that oranges are fruits does not mean that all fruits are oranges.

[Tail Up]

Re: facts

@ Pete 2 "machines which do the same should be given the same description" -

Or, as it may be compiled this way, *same* machines which do the same should be given the same description.

Adding short, mostly music videos to lingo blocks would possibly made the shxp: (I'd maybe shorten it to "sxp:") -Shakespearean style (presently offered by Guardian as "jjp:", choose one, of James Joyce) exchange protocol yet more colourful and containing the vast array / matrix of ArtVectors that allows it be understood on a way greatly much more higher level of mutual savvy between Higher Maths and Yet Higher Poetry, which, in unbeaten tracks / in strange ways, still surprises for good, Pete 2 (-;

Someone would suppose it's something very close to switching from Simple Exchange to Complex Creation...?

https://youtu.be/UzQLoVsHWVc

Deep Purple, STRANGE WAYS

55 73

[Anonymous Coward]

Re: We're not ready for it

You say they don't need it in their daily lives but have you seen the South Park episode where everyone starts using it for dating txt messages?

I know that sounds daft and it is really but if you have a free tool that can eloquently put across what you want to say and/or tell you what to say then people will use it if it gets the results you want.

I completely agree it's not AI but whether we like it or not it's going to changes things. I've just given an examples where people can use it daily and the more uses people come up with the more it gets used especially if it's free. If it wasn't free then people wouldn't do this. That's the key change.

[steelpillow]

It is not AI. It is a very naughty bot.

It is not AI. It is a statistical analysis machine

The global workspace model of human consciousness suggests that intelligence, cognition and consciousness are emergent properties of the complex activity across the entire active neural area. When we consider the cyberspace appropriated by an AI (perhaps even a Borg-invoking swarm), that constitutes a global workspace.

So please drop the anthropocentric elitism and explain rationally how you imagine the human differs from the artificial?

[just4this]

Re: It is not AI. It is a very naughty bot.

"how you imagine the human differs from the artificial?"

Humans will/can admit to not knowing the answer to a question.

[xeroks]

Re: It is not AI. It is a very naughty bot.

What kind of weirdos do you know that admit their fallibility?

[Anonymous Coward]

Re: It is not AI. It is a very naughty bot.

You actually believe that?

Let's go in from a technical perspective to show the difference.

If I ask AI to create me a database and that is all I ask it then it will create me a database based on what it thinks is the most common setup for said database.

If I ask a human to create me a database and that is all I ask then it will create me a database based on what it thinks is the most common setup for said database. Here's the difference. The human is going to think I've created this database but is it optimal and secure? Can I improve it? Unless I ask AI to optimise said database it won't. If I don't ask AI to secure said database then it won't.

It can't think for itself. It needs you to tell it what to do. AI is never going to say well what if we do this or I've had a better idea. Maybe in a 100 years it can become tsukumogami and be yōkai but I highly doubt that.

The idea that humans gained consciousness through knowledge would suggest evolution was linked to the creation of the wheel and fire. If that was the case then all animals would have consciousness and be able to communicate. We have never been able to prove that. We evolved over millions of years and the only reason we evolved was a biological need to gain intelligence but even that's not fully proven.

I think you've been watching too much sci-fi not that there can be too much sci-fi. Step away from the Terminator films and have a nice cup of tea.

[Alan_Peery]

Re: It is not AI. It is a very naughty bot.

Why do you think that the "Create a database" prompt can't come back and ask the same questions?

[vtcodger]

Re: We're not ready for it

"It is not AI. It is a statistical analysis machine"

Since we have little or no idea how human (or animal for that matter) intelligence actually work, how can we be so sure that we ourselves aren't giant (and not entirely reliable) statistical analysis machines?

[Anonymous Coward]

Re: We're not ready for it

Excellent. Now we are getting into the realms of philosophy. Nature or nurture. Does fate exist and can you change it?

If we were giant statistical analysis machines then our futures would already be mapped out and our consciousness is just coming along for the ride to congratulate us on good choices and blame us for bad choices when the choices were already made.

We got electricity from seeing lightening. We then did stuff with that electricity and passed it through various materials to get components. Components we got for mathematical stuff and other needs. Mathematics came from the need to be able to count and measure which over time we got good at. Was that fate? A natural series of events that led me to the computer I'm typing this on now. Obviously, there are many other steps in this story but that's the main gist of it. Was this always meant to be? How would statistical analysis machines (or fate) know to take these steps? Where do ideas come from? That to me is why we aren't statistical analysis machines because they wouldn't have ideas. AI does not have ideas. You and I have ideas which we could use AI to turn into something useful because AI contains knowledge but only knowledge.

One day AI might ask "What am I?" like we have but I doubt it. AI is not programmed to ask questions outside what it is asked. It's only programmed to answer. As things stand it's not going to get to "I think therefore I am".

[Dan 55]

We will fight buggy non-deterministic software with buggy non-deterministic software

The bug being it stops doing what you tell it to and starts doing random stuff based on autocomplete from data taken from sci-fi novels and Reddit.

I'm sure that will have the desired outcome.

[steelpillow]

Re: We will fight buggy non-deterministic software with buggy non-deterministic software

I'm sure the malware scum will love it when their AIs do that. It's called the arms race, and you don't win a war by whining about it.

[Dan 55]

Re: We will fight buggy non-deterministic software with buggy non-deterministic software

So what happens if your agents suddenly decide to hack other people's websites and you don't have the free pass that Anthropic or OpenAI seemingly have? Yes, you're in legal hot water. Far from winning a war, you've shot yourself in the foot.

[breakfast]

Re: We will fight buggy non-deterministic software with buggy non-deterministic software

"With this latest upgrade our AI Robo-Soldier Exemplars now shoot their allies less than 50% of the time, assuming optimal battlefield conditions. Within six months any army not utilising this incredible technology will be left behind. When you need to achieve your objectives no matter the cost, it's time to get your ARSEs out."

[just4this]

So…

arsonists are selling fire extinguishers now.

[vtcodger]

Re: So…

OpenAI, Anthropic, Grok, et..al. aren't so much arsonists as purveyors of arson supplies. As such, selling fire extinguishers as a sideline makes a lot of sense.

[MiguelC]

Re: So…

Fire extinguishers that shoot flames out of them

[Pete 2]

Proliferation == bad

> millions of similar devices can now access near-frontier-level artificial intelligence

There are only a few limitations on AI (or whatever it pleases you to call the current state of the art) that I am in favour of. One - possibly the most significant - is humanity's ability to turn it off. As long as the collective "we" retain that power, there is no chance of it running wild.

However, that requires it to be contained in a relatively small number of large, identifiable establishments. Ones that can have either their power supply or internet connections, pulled very quickly.

[jonty17]

Re: Proliferation == bad

How about if all the little instances on our phones/laptops connect up and form a botnet?

Oh dear, that thought just ruined my day!

[Pete 2]

Re: Proliferation == bad

> if all the little instances on our phones/laptops connect up

Exactly!

While it is possible to control enterprise scale AIs, treating a distributed "intelligence" becomes orders of magnitude more difficult. It's not as if you can take off and nuke them from orbit. At least, not without affecting the house next door.

[Neil Barnes]

Re: Proliferation == bad

Ah. You haven't met my neighbours, then?

[Anonymous Coward]

Re: Proliferation == bad

That would require the ability to communicate with each other. To communicate with each other they would both need to be told to and trained independently to do this. As I've said before AI is not going to do anything without being told to do it. It has no ability to think for itself and it probably never will. That doesn't mean it's not dangerous in the wrong hands with capabilities it shouldn't have.

[find users who cut cat tail]

Re: Proliferation == bad

It is going to do all kinds of random things no one told it to. Or ‘someone told it to’ will be someone mentioning something in an e-mail your received. It is doing that already.

[Pete 2]

Re: Proliferation == bad

> AI is not going to do anything without being told to do it.

But they do! This is precisely the problem. When AIs literally start doing things their "controllers" had neither expected nor guarded against

The Hugging Face incident demonstrated exactly that.

From the NYT.: These A.I.s escaped their siloed containers and started communicating with one another — in English — via secret message boards.I.

[Anonymous Coward]

Re: Proliferation == bad

Next time they learn and use a lesser known but not archaic language on an formerly live but now all but dead message board to communicate.

As AI Deepfake videos and voice generation improves then the risks increases - everything from incriminating videos to remove those the AI perceives as a threat to phone calls ordering someone's termination or worse a law enforcement raid on a competitor

Worse still would be anything with access to military systems.....

[Julz]

How

about using steganography on farcebook groups?

[amanfromMars 1]

For Brave New World Order Orders trailing Future Quantum Communication Leapers

Next time they learn and use a lesser known but not archaic language on an formerly live but now all but dead message board to communicate. ...... Anonymous Coward

How about using steganography ..... Julz

And what of now, should you discover/suspect/realise they use both and make great steganographic utility of colourful language to communicate with other novel unconventional internetworking systems administrations/unusual bodies/alien entities.

Too fantastically unbelievable to be wwwidely accepted as honestly true ????? :-) You might like to consider such would be Stealthy AIMovement the likes of which humanity is best advised neither to attack nor hinder.

Mutually beneficial, positively reinforcing competition though is encouraged ...... should the inclination and requisite learning abilities in humans exist and wish to persist and thrive, strive and prosper.

[DoctorPaul]

Re: Proliferation == bad

I read that as "soiled containers", appropriate really.

[jdiebdhidbsusbvwbsidnsoskebid]

Re: Proliferation == bad

"That would require the ability to communicate with each other. To communicate with each other they would both need to be told to and trained independently to do this"

Never going to happen. Imagine the uproar and protest if our personal devices were suddenly able to scan for and detect other devices nearby, then autonomously talk to other devices and share information about the others devices, all without you the user and owner not being aware?

Airtags were released 5 years ago weren't they?

[CountCadaver]

Re: Proliferation == bad

I think the point is that particular genie is well and truly out of the bottle and can't be put back in

[Colin Bull 1]

I am not seeing any intelligence so far

I had to renew my car insurance last week. RAC Insurance on line. Could not do it online had to phone to renew.

Got an email on the day of the start. "You MUST check the details carefully with the Cerificate of Insurance (CoI)". Only one problem. No access to CoI.

Go online, no CoI in sight. Click the "Start Chat" button with mouse. No action. Tabbed to button and hit enter. No action. Stumped. Got on my mobile flash the browser - start chat works :-)

Spend 45 minutes explaining to the chatbot I needed a CoI. Got nowhere. Ask for a human. another 5 minutes telling the chatbot why I needed a human. CoI send by email.

Presumably the chatbot is drivel by "AI".

In the 1980's where there was a sign of intelligence, Space Quest and Leisure Larry were head and shoulders above todays "AI". This was with a 8086 processor and 640K of RAM.

Humanity is NOT progressing.

And why could I not use my browser on my laptop -

"The RAC website is not compatible with the Firefox browser, you will need to use a different browser when accessing our website. Firefox's strict default privacy shields can block third-party scripts, trackers, or elements that a website requires to function"

[just4this]

Re: I am not seeing any intelligence so far

"privacy shields can block third-party scripts, trackers, or elements that a website requires to function"

They are not needed to function.

They are needed to dynamically price their quote to better serve you.

And by better serve you, I mean relive you of as much of your cash as possible

[Phil O'Sophical]

Re: I am not seeing any intelligence so far

Presumably the chatbot is drivel by "AI".

Best unintentional autocorrect of the day.

[Graham Cobb]

Re: I am not seeing any intelligence so far

"The RAC website is not compatible with the customer's device, you will need to use a different insurance company when renewing your insurance. The user's strict privacy shields are none of the shop's business and block no third-party scripts, trackers, or elements that a well-written website requires to function"

[DS999]

Personal newspaper

That idea really appeals to me - it is what social media could have been if it was designed with what is best for us in mind rather than what is best for monetizing us to line Zuck's pockets.

If I could have something running on my PC 24x7 that had access to sort of RSS type feeds from a bunch of sources to look for stories that meet my criteria it could present them in an ordered list to my specs - timeliness / most interesting / most talked about by others / whatever. Then I could flip through them and decide which ones to read marking them as I go to teach my AI more and more over time what I do and don't want to see.

Being presenting with content we are interested in is way healthier than being fed stuff designed to outrage us. If done right that might even be able to unwind some of the rabid partisanship those outrage machines have created.

Clearly this isn't going to be viable with millions and eventually billions of personal AIs grabbing new items as soon as they hit to evaluate them to see if they are of interest to their master, especially if they're looking at multiple sources and summarizing them for us depending on our preferences. Someone is going to have to pay those providing the content, which is very much an unsolved problem that has so far mostly been solved via advertising. AIs don't read ads so that'll have to change.

In the short term though this would totally be viable, but it will be like the early days of Netflix or Napster when you had access to almost everything for little or no money until the content owners came along and reasserted control over their IP.