proposal

At least make any AI images identifiable as such.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

already an EU requirement, AI content must be marked as such and have digital watermarks to identify the AI source, though I think some parts of it has had deadlines for compliance extended

the AI companies want that too because they don't want to train their AIs on AI generated content

I've noticed that many Youtube videos have smothed faces, as if a lot of makeup had been applied. Weird.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Right now they are only smarter in a few distinct niches, but they are not all that far away from surpassing the best humans in more general intelligence. Right now a well behaved high end AI can mimic a graduate student pretty well and even a post graduate researcher in some domains.

It has come on a long way in the past 5 years. One side effect is that high speed large ram modules are now ridiculously expensive to buy.

Most of them still are to the cognoscenti. Facebook even tags some harmless content that *is* AI generated correctly as "AI generated".

Antisocial media is hungry and ripe for rage baiting and the algorithms find that is the most profitable thing to do and so AI's and malevolent humans are happy to provide such content and bank the click through fees.

The problem is that most people are so gullible. That is why social engineering can beat even the most sophisticated security systems.

What do you expect that to achieve?

The genie is out of the bottle. How is your proposal going to prevent me (or any other actor) from using a *local* copy of an open source/weight AI and continuing its training, indefinitely? You've saddled "public" AIs but can do nothing to stop "progress" that happens outside of your sight.

I suspect the concensus is that cloning a human being should be prohibited -- as immoral. Yet, I'd lay batter than even money that there's just such a clone *somewhere*!

What do you expect that to achieve?

The genie is out of the bottle. How is your proposal going to prevent me (or any other actor) from using a *local* copy of an open source/weight AI and continuing its training, indefinitely? You've saddled "public" AIs but can do nothing to stop "progress" that happens outside of your sight.

I suspect the concensus is that cloning a human being should be prohibited -- as immoral. Yet, I'd lay batter than even money that there's just such a clone *somewhere*!

Too subtle for you, but I consider myself smarter than humans too. It would be great for long cons and felony fraud.

Op 20-9-2026 om 22:27 schreef Lane W:

We should actually speed up AI development. Current AI is garbage. It can't even come up with concepts from scratch for toy problems like the game of go.

The biggest existential threat to humanity is mass stupidity and AI is our best bet to solve that problem so that humans may develop in more sustainable and harmonious ways.

AI can fix capitalism, so we actually get quality products for a very low price or free instead of fake capitalism and getting ripped off by overpriced garbage.

AI can fix democracy so we get rid of clueless politicians that violate human rights and replace them with an AI government that is able to secure human rights for everyone where everyone is able to participate in the government with an influence proportional to their level of understanding. To ensure that people are all aligned globally towards mutual prosperity.

Our entire economy is bogus and based on scams like intellectual property (similar to other scams like slavery in the recent past). Everything that people exchange online (music, books, video, pictures, apps, etc..) is just a bunch of numbers and all numbers belong to the public domain. People who think numbers can be owned as intellectual property belong in a mental institution.

Mankind has been bent on omnicide for at least 10,000 years that we know of, and they're still here. All the doomsday publicity is just that: a publicity stunt.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Slightly less than half of the population can make that claim so it's a low bar to cross.

There will be all sorts of less obviously malicious consequences when people discover they can deploy their own agents.

When the local grocery store realizes they can charge *you* different prices from the guy standing NEXT to you. When the fast food outlet realizes they can raise prices by a small amount at certain times of the day without irritating their customers enough to cost them sales.

What will be MORE interesting is to see when consumers realize they can similarly leverage those abilities to find the best prices, vendors, products, times to purchase, etc.

A technology on a par with nuclear but at an affordability (and ubiquity) of gunpowder. It will be interesting to review early advocates' predictions against the reality. And, the contortions of AI vendors and gummits in trying to put the genie back in the bottle (like the British trying to ban the harvesting of salt to protect that market...)

Fun times!

"Surely the days of programming, as we know it, are numbered ... Instead, we’ll express our intention about what should be done... Then these expressions will be submitted to immense, intelligent, intention-understanding programs, which will themselves construct the actual, new programs."

-- Minsky ca 1983

Sad that he missed it all by a decade (but, then again, that may be fitting as AI was always "a decade away")

Of those you mention, music takes notoriously more numbers to represent, with video following second. Are we just waiting for the next media breakthrough? What's it going to look like? Something that the current numbers can't represent or take a decade to catch up on. Personally, I think music's not quite captured yet. My streaming music app is the least predictable function on my computer in many ways. Takes the longest to load, costs 3 times the basic service, and still has numerous holes in availability.

Remember ADA? "It is symbolic that the last programming language is named after the first computer programmer, Ada Lovelace"... "Does anybody here remember Vera Lynn", duh...

Those of us who were in the industry when computers were BIG lived through numerous breaktroughs, firsts and lasts, AIs (remember Lisp/Prolog and whatever else? Perceptron anyone?) and much more. The main difference between us, old farts, and the new generations is that we DO remember the previous revolutions that fizzled. The new generation re-discovers those shiny new things that they don't know already existed, fizzled, and got forgotten... Anybody remembers "There is an app for that" from several decades ago?

In this industry, old ideas are continually recycled.

Big machines -> time sharing -> personal computers -> X terminals -> edge computing -> centralized AIs -> AI-enabled phones, etc. What was once old "suddenly" becomes new (and NOVEL!). Remember Motogorilla's ICU??

But, often "old" simply failed because it was "ahead of its time". Or, not economically viable /at that time/. In the early 70's, we used i4004's instead of dedicated logic -- and that was "revolutionary" ("We can CHANGE how it behaves without having to shoot a new PCB!"). Now, I use gigahertz multicores instead of dedicated logic for similar reasons -- just at a different scale (and price point).

Athena was "revolutionary" in giving each user a VAX! Now, that's a big yawn; your phone has more /local/ resources and costs a tiny fraction! (but, we'll treat it as an access point to a big AI cluster, going forward -- until we can migrate that cluster back into the phone...)

The one constant is that software just keeps getting applied to more and more problems so any "advances" disappear in an ever-increasing field of applications. ("Gee, we can write a piece of code to implement a bunch of logic functions. Or, a radio!")

And, each of those domains requires even more *tools* to apply those technologies (which beget other applications which beget...). Talk about job security!

Do you really need an AI to do some simple asm programing to do some -sometimes not so simple- things? I do asm all the time in PIC micros:

There are thousands of programming languages these days:

Essence in my opinion is: know your hardware.

In the sixties of the previous century humans could do a manned Moon return.

And now with all the ships, sorry Sjips, ohh, chips??? and a bit of high altitude cosmic particles .. ever get somebody to the Moon and back?

I am not against AI, but then... playing tennis against a computer? Who will be not allowed in the Olympics: humans or robots? Do NOT underestimate human brain power, human beings doing things.

Next generation ? even reading and writing is in decline for the kids I've read. OK Rosetta stone was difficult too, but... Language next? brain electrodes to commi-nukate eeeh communicate? Bye Bye species?

Sorry it is still early here... LOL

AI creating humans got billions from AI jive on the stockmarkets, and now want to relax and spend the money on nice food and way of living.

The US war house dissident wants AI to continue to make more money from it. He needs more AI to help him kill humans.

I have never programmed PICs in anything but assembly. And I did it hundreds of times.

It was OK when MPLAB still had MPASM. Then gputils with their gpasm came to life and MPLAB was no longer needed/used at all.

The new one is definitely something out of old "Toster" parable -- now they force windomorons to program everything in C. Generating relocatable object files and then linking them into a final binary. Even for the tiniest chips with 2 level deep hardware stack, 256 words of program memory, and 16 bytes of RAM. Very clever decision and absolutely necessary, duh... Those should be programmed in C, duh...

However, I stopped using their MPLAB long ago, when gputils were born. The CVS version is significantly behind and does not have support for newer chips but I maintain my own version where I add the nice late ones I use. That means I still program those in Assembly and that is still good old MPASM, not that horrible weirdo they have in their MPLAB (whatever it is called these days) now. The gpasm just works and produces the final image from MPASM source.

Vera Lynn lived in Ditchling Village in Sussex, which is where my wife lived when she first started work at Sussex University in Brighton.

Then again there is the protein folding problem that Google's Deep Mind solved a few years ago, and the formalisation of the proof of Fermat's last theorem. Stuff is happening, but not where the "head up their own behind" programming community is paying attention.