Schneier on Security

Clive Robinson • October 11, 2024 4:25 PM

@ Bruce,

I would not go with titles 2&3, 1&4 are more open to change that will happen when you are writing the book.

Also,

“What we want out of the title is that it convey (1) that it is a book about AI, (2) that it is a book about democracy writ large (and not just deepfakes), and (3) that it is largely optimistic.”

As the old joke punchline has it

“You can have any two out of three”

The one you won’t get is the third.

Politics is about power, and power is about discrimination and theft along with other harms to those who lack power.

AI of any form, not just the current LLM/ML fetish is a tool, and tools are agnostic to use.

It’s the “Directing Mind” that decides the use, and supposedly impartial observers that decide if the use is as far as society is concerned “Good or Bad”.

The general societal concern about technology is as it has been since the early 1990’s and what the fall of the Berlin Wall revealed was happening in the CCCP states. Much as it still is in Belarus and Hungary with one supposedly on either side of the EU and Russia sphere of influence boundary. But the reality both effectively run as “strong men” dictatorships held in place under Putin influence.

The result is people are aware of the “survailance state” in ways they formerly were not, and importantly the key factor technology plays in it.

You only have to look at Robo-Debt and similar to see how “political mantra” can be put in place with technology with few people involved to devastating effect.

AI will reduce the numbers involved whilst increasing the ability of those “strong men” to not just put in place their mantras but plunge them knife like into the very heart of society. All whilst maintaining an apparent “arms length distance”.

Thus those who should be named and shamed will have a defence of “the computer says” or similar to walk away unblamed and without consequence.

If you doubt this, reflect on what is happening on the US south boarder currently and what is turning into a battle ground driven by political mantra not rational thought. Now imagine boarder patrol by heavily armed drone carrying out “wish fulfilment” for certain “Congress Critters” and similar… If you doubt this there are a couple of other places in the world to think of right now where such things have got to the point of significant bloodshed, and worse, all because “individuals” who prance around as “strong men” and are shoring up their political position on piles of corpses strewn around vast areas of destruction.

Then turn your attention to the daily events in and around the South China Seas area and consider what is happening there and how.

What about the use in China of technology to “surveille the masses” and “weed out” by “social scoring” those that are seen as not being suitable citizens.

This is what AI will bring to Politics and Law, because they are the routes of power. The fact corporations will gladly go along with it has already been painted “writ large” as a warning to all.

Is there “good” that AI can do, yes it can take drudge and human limitations out of creating new. I’ve been involved in one way or another in AI since the 1980’s for “Engineered Solutions” to solve problems. AI is not just LLM/ML systems, it’s got way greater scope, few realise just how much it impinges on their daily lives and how much more it can and will do.

But we in turn have to separate how we are going to apply it, to assist or harm society and the individuals within it.

We all laugh at the notion of an out sized Mall-Cop on a Segway or similar or even those single wheel almost hoverboard devices. You will find that they all have within them a form of AI that had it’s roots in the 1980’s ensuring that those on them can use them. Trains, Planes and Ships likewise use similar work that came from AI to run the semi-autonomous systems that they could not be built and function without.

That is where the “optimism” is and has been getting on toward half a century.

The current LLM/ML systems are going to be very short lived, they are actually hitting the point where the curves are nearly flat thus hitting an effective end stop. No matter how much money you throw at them we are well into “The law of diminishing returns”. Where each minor incremental improvement is purchased at exponentially increasing cost.

Worse they are choking themselves on pollution, both of their output getting back into their input, but their environmental pollution from excessive use of power and water.

I suspect that by the time your book gets to where it would be on the book stand, the current AI bubble will have come to a point where it’s end is fairly clear. Because I doubt there will be a “technology breakthrough” in that time available to save it in the ways required.