AI and Lossy Bottlenecks

Artificial intelligence is poised to upend much of society, removing human limitations inherent in many systems. One such limitation is information and logistical bottlenecks in decision-making.

Traditionally, people have been forced to reduce complex choices to a small handful of options that don’t do justice to their true desires. Artificial intelligence has the potential to remove that limitation. And it has the potential to drastically change how democracy functions.

AI researcher Tantum Collins and I, a public-interest technology scholar, call this AI overcoming “lossy bottlenecks.” Lossy is a term from information theory that refers to imperfect communications channels—that is, channels that lose information.

Multiple-choice practicality

Imagine your next sit-down dinner and being able to have a long conversation with a chef about your meal. You could end up with a bespoke dinner based on your desires, the chef’s abilities and the available ingredients. This is possible if you are cooking at home or hosted by accommodating friends.

But it is infeasible at your average restaurant: The limitations of the kitchen, the way supplies have to be ordered and the realities of restaurant cooking make this kind of rich interaction between diner and chef impossible. You get a menu of a few dozen standardized options, with the possibility of some modifications around the edges.

That’s a lossy bottleneck. Your wants and desires are rich and multifaceted. The array of culinary outcomes are equally rich and multifaceted. But there’s no scalable way to connect the two. People are forced to use multiple-choice systems like menus to simplify decision-making, and they lose so much information in the process.

People are so used to these bottlenecks that we don’t even notice them. And when we do, we tend to assume they are the inevitable cost of scale and efficiency. And they are. Or, at least, they were.

The possibilities

Artificial intelligence has the potential to overcome this limitation. By storing rich representations of people’s preferences and histories on the demand side, along with equally rich representations of capabilities, costs and creative possibilities on the supply side, AI systems enable complex customization at scale and low cost. Imagine walking into a restaurant and knowing that the kitchen has already started work on a meal optimized for your tastes, or being presented with a personalized list of choices.

There have been some early attempts at this. People have used ChatGPT to design meals based on dietary restrictions and what they have in the fridge. It’s still early days for these technologies, but once they get working, the possibilities are nearly endless. Lossy bottlenecks are everywhere.

Take labor markets. Employers look to grades, diplomas and certifications to gauge candidates’ suitability for roles. These are a very coarse representation of a job candidate’s abilities. An AI system with access to, for example, a student’s coursework, exams and teacher feedback as well as detailed information about possible jobs could provide much richer assessments of which employment matches do and don’t make sense.

Or apparel. People with money for tailors and time for fittings can get clothes made from scratch, but most of us are limited to mass-produced options. AI could hugely reduce the costs of customization by learning your style, taking measurements based on photos, generating designs that match your taste and using available materials. It would then convert your selections into a series of production instructions and place an order to an AI-enabled robotic production line.

Or software. Today’s computer programs typically use one-size-fits-all interfaces, with only minor room for modification, but individuals have widely varying needs and working styles. AI systems that observe each user’s interaction styles and know what that person wants out of a given piece of software could take this personalization far deeper, completely redesigning interfaces to suit individual needs.

Removing democracy’s bottleneck

These examples are all transformative, but the lossy bottleneck that has the largest effect on society is in politics. It’s the same problem as the restaurant. As a complicated citizen, your policy positions are probably nuanced, trading off between different options and their effects. You care about some issues more than others and some implementations more than others.

If you had the knowledge and time, you could engage in the deliberative process and help create better laws than exist today. But you don’t. And, anyway, society can’t hold policy debates involving hundreds of millions of people. So you go to the ballot box and choose between two—or if you are lucky, four or five—individual representatives or political parties.

Imagine a system where AI removes this lossy bottleneck. Instead of trying to cram your preferences to fit into the available options, imagine conveying your political preferences in detail to an AI system that would directly advocate for specific policies on your behalf. This could revolutionize democracy.

a diagram of six vertical columns composed of squares of various white, grey and black shades

Ballots are bottlenecks that funnel a voter’s diverse views into a few options. AI representations of individual voters’ desires overcome this bottleneck, promising enacted policies that better align with voters’ wishes.
Tantum Collins, CC BY-ND

One way is by enhancing voter representation. By capturing the nuances of each individual’s political preferences in a way that traditional voting systems can’t, this system could lead to policies that better reflect the desires of the electorate. For example, you could have an AI device in your pocket—your future phone, for instance—that knows your views and wishes and continually votes in your name on an otherwise overwhelming number of issues large and small.

Combined with AI systems that personalize political education, it could encourage more people to participate in the democratic process and increase political engagement. And it could eliminate the problems stemming from elected representatives who reflect only the views of the majority that elected them—and sometimes not even them.

On the other hand, the privacy concerns resulting from allowing an AI such intimate access to personal data are considerable. And it’s important to avoid the pitfall of just allowing the AIs to figure out what to do: Human deliberation is crucial to a functioning democracy.

Also, there is no clear transition path from the representative democracies of today to these AI-enhanced direct democracies of tomorrow. And, of course, this is still science fiction.

First steps

These technologies are likely to be used first in other, less politically charged, domains. Recommendation systems for digital media have steadily reduced their reliance on traditional intermediaries. Radio stations are like menu items: Regardless of how nuanced your taste in music is, you have to pick from a handful of options. Early digital platforms were only a little better: “This person likes jazz, so we’ll suggest more jazz.”

Today’s streaming platforms use listener histories and a broad set of features describing each track to provide each user with personalized music recommendations. Similar systems suggest academic papers with far greater granularity than a subscription to a given journal, and movies based on more nuanced analysis than simply deferring to genres.

A world without artificial bottlenecks comes with risks—loss of jobs in the bottlenecks, for example—but it also has the potential to free people from the straitjackets that have long constrained large-scale human decision-making. In some cases—restaurants, for example—the impact on most people might be minor. But in others, like politics and hiring, the effects could be profound.

This essay originally appeared in The Conversation.

Tags: artificial intelligence, LLM, voting

Posted on December 28, 2023 at 7:01 AM • 44 Comments

Comments

Clive Robinson • December 28, 2023 7:31 AM

@ Bruce, ALL,

Re : The flip side of choice.

“Today’s computer programs typically use one-size-fits-all interfaces, with only minor room for modification, but individuals have widely varying needs and working styles.”

The reaaon for the apparent “one size fits all” is effectively two fold from complexity.

Increased choices requires increased complexity that grows faster than the choices…

One reason for this is a data object becomes available to a greater number of methods. However not all methods can or do work the same way. For instance a nibble can be used to store a hexadecimal digit, a “Binary Coded Decimal”(BCD) digit or a “Signed octal” digit. You do not know looking at the nibble contents what type of data is stored and more importantly how the basic methods to manipulate it are supposed to work. Thus you get ambiguity, for instance if the value in the nibble is 0xA, is it,

1, A hex digit
2, An error on BCD
3, A signed Octal digit
4, A partially compleated method result such as adding two digits but not yet finished resolving the carry.

Which leads on to the second major issue, as a rough rule of thumb as choices increase, complexity increases and a large number of securiry issues can and do arise.

Thus an argument can be made that as LLMs, ML etc AI increases choice, the security not only decreases but the non atomic nature means that it’s not just vulnerabilitires that increase complexity for checking has to increase as well. At the simplest the increase goes up by the number of interactions between objects thus as N^2…

Rene Bastien • December 28, 2023 8:20 AM

Very interesting. I do have questions. As a societal level as well as a personal level, how can we ascertain that such a system would effectively convey our opinions? And how would we be able to modify our opinions and beliefs in time?

Beatrix Willius • December 28, 2023 8:21 AM

Why should this be a desirable future? It sounds dystopian instead. We already have too many choices – except in the Murican party system.

Iskra • December 28, 2023 8:32 AM

Had same idea for a while now…great that it starts to penetrate some… To make this a reality, we have to make this a new religion and convert everyone. With all politicians against this, no other law to protect but religious freedoms. Same will allow to opt out of existing voting rituals into a new one. Have Elon provide the social backend and we have created something new, not something designed during era when information exchange was handicapped by speed of horses.
DDD= Demand Direct Democracy, with AI as the ruler aiming at the voted results. Vote for results, majority wins. Once aims are set, AI invents a method to achieve, given moral and law input guardrails. Go go go humanity.

Commenter n+1 • December 28, 2023 10:02 AM

As a followup to one of the previous questions, how would the mapping between personal beliefs and voting choice be determined? Consider that today many people actively vote against their beliefs or best interest primarily due to outside influence or misinformation on what any particular vote means. If the system votes on a person’s behalf using a “correct” mapping, that person may believe that the system voted incorrectly on some subset of votes.

yet another bruce • December 28, 2023 10:44 AM

Very thought-provoking, thank you.

AI-mediated meal planning sounds interesting. I am not sure that the main purpose of limiting menu choices is to simplify communication and decision-making. Another important function is streamlining kitchen operations and staff training. An AI-mediated bespoke menu selection could easily turn into an episode of Iron Chef at the back of the house.

It could be a challenge to maintain trust in AI-mediated political decision-making. AI systems based on machine learning currently appear to be black boxes which undermines trust. This is particularly true if the AI output is unexpected. Something like AlphaGo’s famous “Move 37” could cause a lot of consternation in a political context.

JPA • December 28, 2023 10:46 AM

I think this would lead to increased personal distress. If someone wants to live happily, then one of the most important qualities they need to develop is the ability to be content when they don’t get exactly what they want. Having to compromise and prioritize some desires over others develops this quality.

If a person is used to getting all their desires met then how will they be able to deal with situations in which that can’t happen because of physical constraints, or the refusal of other people to give them exactly what they want?

Douglas K • December 28, 2023 11:28 AM

“Today’s streaming platforms use listener histories and a broad set of features describing each track to provide each user with personalized music recommendations.”

And they are astonishingly bad at it.. first they will play the music you already listened to, then the most banal and obvious similar tracks, then go back to playing the music you already listened to. Radio stations still are a better choice for finding new music. A good human making a playlist will outperform any of the digitally extruded lists. Any public radio station in a university town is a good place to start.

More generally the problem here is the assumption that we already have AI that can do wonderful things. To pick on just one example,
“By capturing the nuances of each individual’s political preferences in a way that traditional voting systems can’t”
This is handwaving. Current LLMs can’t do this, and I don’t see any mechanisms proposed that would be capable of it.

I agree, if we had that wonderful AI many things might be possible. Unfortunately what we have is LLMs capable of text extrusion, carefully hand-built on top of IP theft and vast numbers of poorly-paid humans labelling and categorizing the stolen property.

JonKnowsNothing • December 28, 2023 11:55 AM

@All

A few comments:

Who ever decided to use a restaurant metaphor never worked in a restaurant

Yeah your Mom might have made you your favorite dish because she knew it was your favorite and kept supplies in a Just For You supply chain.

Restaurants, dinners, fast food, taco trucks and eateries of all kinds including bakeries, markets, meat markets, farmers markets do not work like this example

Multiple Choice Voting

Ranked Voting has been around for a long time. There are places already with ranked voting and doing very well with it. It exists in RL and it is common in many games where elections are part of the game play.

The term ranked voting, also known as preferential voting or ranked choice voting, pertains to any voting system where voters use a rank to order candidates or options—in a sequence from first, second, third, and onwards—on their ballots.

It works. It is not an invention of AI-HAIL Stormers

Complex Consumer Choice

The biggest problems with consumer choice is:

Fear of missing out
Fear of getting what is not enjoyed

The rest of the problem is financial. If you have the funds for a Maserati, you are not going to worry much about buying it. No AI is going to help buy one if you haven’t got the credit score to get a loan for it.

Proliferation of AI Stochastic Parrots

Ads and puffery have been around since someone offered a basket of wheat for a basket of apples. Currently, people know to check the contents of the Pig in a Poke most of the time.

AI adds nothing to the poke and nothing to the basket other than tell you it’s a basket of “highest quality items to be found anywhere…”.

Humans still have to figure out how much AI is puffing up whatever might be in the poke.

A MSM article indicates that in the UK, a long standing newspaper will be publishing “AI Assisted Reports” in order to free up journalists to find “real news”. Essentially declaring they are changing into the National Enquirer.

Will you be able to tell a Journalist’s Martian from an AI Martian?

Matt • December 28, 2023 12:17 PM

You need to think more like Cory Doctorow. The question isn’t “what fantastic new applications will AI give us,” the question is “how will the powerful use this tool to continue abusing everyone else?” You suggested “what if employers could look at all your coursework in detail” and my first thought was “Why in the actual f*** would I ever want an employer delving into my past that deeply? Get out of here with that line of thinking. Sometimes a lossy bottleneck has social value: I can’t leave my past behind, I can’t change, if everyone else in society always has access to absolutely everything I’ve ever done in detail.

Clive Robinson • December 28, 2023 12:52 PM

@ JonKnowsNothing, ALL,

Re : Fast food is rarely “cooked”.

“Restaurants, dinners, fast food, taco trucks and eateries of all kinds including bakeries, markets, meat markets, farmers markets do not work like this example”

Most of them “do not cook ingredients to make food”. They generally “buy-in” finished items and heat them through, and maybe turf on a bit of mixed salad from a nitrogen bag etc.

Even high class restaurants work in an industrial way.

They actually have a very limited set of ingrediants they purchased in and they are cooked seperately.

The method is sometimes called,

“Protein, Carb, Veg and sauce on top”

It’s primarily done to save wastage.

Put simply the protein is cooked in one of a very few ways such as grill roast boil and is dumped on the plate and a sauce/gravy sloshed on to give the supposed dish. The carbs will be potato, rice, or wheat based like pasta, cooked in large pots on continuous rotation. The veg will be cooked likewise but in much smaller amounts you get maybe three of four and will have a sauce sloshed on if it’s mostly green leaf. Sauces are made similarly a flour, fat, cooked out will have milk or plain stock added, for your dish they will chuck in pre-cooked mushrooms, herbs, etc wihisk in butter or liquid egg then quickly heat it through to get the flavour from the herbs and glos from the buter egga and slosh it over. The basic meat, fish and vegtable stocks are often “out of a can” or tetra pack probably bought in bulk months before.

You can do something like sixty supposadly different dishes this way with little difficulty.

If you want to see it taken to the almost ultimate way have a look in “UK/US Indian Restaurants” they basically make hugh vats of “masala” gravy of tomatoes and onions with a basic masala ginger base flavour and spices. To make a curry they put about a tablespoon of oil in what looks to many like a wok on an intense heat and when it’s shimmer hot add in half of the other spices. After a moment or so they throw in a portion of often pre-cooked protein and sear it if required orherwise a large ladle of masala is added and the other half of the spices added on top to go from masala to phaal (one of a number of “invented in Birmingham” indian dishes that amazingly have been exported to India and surounding regions),

https://www.kitchensanctuary.com/curry-base-gravy-sauce/

It’s quick, efficient and minimizes waste.

fib • December 28, 2023 1:12 PM

This is a case where the famous Clive Robinson’s dictum apply: technology cannot solve a social problem [because it is not a problem, to start with]

Technology is about convenience. Call me luddite but things like voting [and perhaps praying] should never be object of technological convenience schemes.

Anonymous • December 28, 2023 2:13 PM

Some of the examples of AI in this post are incoherent and seem to be positing mind-reading and not a very well-thought-out variant of it at that.

How are you to know your favorite food without trying new foods? Your favorite clothing without giving some styles a try out in public and seeing the reaction of others?

Employers using AI to further weed out who to hire and not hire sounds dystopian and, frankly, super stressful to be subjected to.

And user interfaces customized to people’s needs? What world are you living in? User interfaces are customized for engagement and sponsorships/advertising, user needs and preferences be damned.

That whole section of the writeup posits mind reading technologies and a vastly different set of business incentives which don’t rely on maximizing revenue per user.

Paul Lock • December 28, 2023 4:55 PM

One word – verifiability.

StephenM • December 28, 2023 6:46 PM

@JonKnowsNothing

“Who ever decided to use a restaurant metaphor never worked in a restaurant”

Exactly.

There are time restraints in the kitchen. Food has to be presented to a good standard at the right price so that diners keep coming back. The staff has to be paid.

Food has to be ordered in in bulk in advance. The cool room, dry store and freezers must all be properly stocked according to what is on the menu.

Prior to service extensive preparation has to be done. Mise en place must be done for each section in the kitchen in the right quantity.

Tables will have different dishes from different sections. All must come together at the same time. Often one dish will have contributions from different sections. Each cook must know all the dishes of his/her section. There is no time to read recipes during a busy service.

google is jealous of tik tok • December 28, 2023 9:33 PM

All I see is “A.I.” being shoved into a rebooted American slavery role.
The premise that “A.I.” = “techno armageddon against human(oid)s” I cannot believe.
In fact, I’m allergic to those kinds of claims.

Meanwhile, in terms of bizarre security reality, what’s CERN got to do with vacuuming up too much of everyone’s internet based metadata? Since pseudopublicly CERN claims authorship of the world wide web (www, in 1989, right?), then what can we expect from people who have no qualms about some of the most obscure uses of hyperexpensive technology that most everyone else worldwide doesn’t need and doesn’t seem to want?

In terms of weapons of mass destruction research (at CERN), what’s that got to do with all this online gossip about [everything that ever flows on the world wide web]?

Last but not least importantly, what’s all the internet porn and it’s bizarre causes and effects and profiteers got to do with online security or it’s lack in the context of the aforementioned CERN and their fetishistic obsession with “paraphiliac” forms of astrophysics?

Really, this is what I want to know.
Any answerers?

Winter • December 28, 2023 10:43 PM

@google is jealous of tik tok

Since pseudopublicly CERN claims authorship of the world wide web (www, in 1989, right?),

You are contesting a history probably half the people here saw unfolding real-time.

What is the point of your incoherent comment? A hatred of High Energy Physics? Or just European HE Physics? Or science in general?

Clive Robinson • December 28, 2023 10:57 PM

@ Rene Bastien,

Re : Preferences against time.

“As a societal level as well as a personal level, how can we ascertain that such a system would effectively convey our opinions?”

Put simply “you can not”.

The only way you could try is by “testing”. The problem as car owners doscovered is that the system can detect it’s being tested and modify it’s behaviour to suit.

“And how would we be able to modify our opinions and beliefs in time?”

Again you can not because the system can be gamed by a person in control of the system. That is they create a situation where you and others have to vote faster than you can update your preferences.

This game is played in some “two party” elections, where the current leader of the party in power selects the date of the next election.

A variation on that is gerrymandering where you redraw the electral boundaries to give you the best odds of being elected.

Then there is changing voter eligability rules very close to the election so you stop those who might vote against you being able to vote and string out the re-register process so they can not be registered in time to vote.

In short you have to assume as a basic tenent of these systems that they are all “evil by default” and try to work out how to stay ahead in the game, which is always a “Red Queens Race” at best.

Garabaldi • December 29, 2023 12:35 AM

Clearly the AI’s have gotten much better and are trying to take over, starting by replacing influencers and attempting persuasion.

We need to plan a rescue for Bruce.

bl5q sw5N • December 29, 2023 1:02 AM

@ JonKnowsNothing

Ranked Voting

Donald Saari in several books analyzes the “difficult geometries” implied by many commonly mentioned vote balloting and tallying ie aggregation methods, see e.g. [1].

His mathematical result is that all ballot-aggregation methods are subject to outcome paradoxes, but that the Borda Count, a particular “pure” ranked vote method, fares the best.

Other ranked vote schemes and serial rank voting plus instant runoff/vote transfer do less well.

A lot, perhaps all, of the advocacy for these non Borda Count ranked choice voting schemes seen on the web reads like “one true method” frenzied enthusiasm.

https://maa.org/press/maa-reviews/chaotic-elections-a-mathematician-looks-at-voting

Ron Helwig • December 29, 2023 4:47 AM

This all sounds very dystopian. It can end up amplifying government, which is already too invasive and controlling.
Instead we should be looking at ways to use AI to reduce the desire for government.

Winter • December 29, 2023 6:22 AM

@Ron

Instead we should be looking at ways to use AI to reduce the desire for government.

There is this eternal desire for lowering or eve abolishing taxes in the US.

Whenever Americans act on this desire, the results are “natural” disasters that burn down or flood cities and kill many. People are poisoned by unsafe tapwater or die by the million of preventable diseases.

But the desiriis unbroken.

Clive Robinson • December 29, 2023 6:41 AM

@ Ron Helwig, ALL,

Re : If it’s, not governed it’s disordered.

“Instead we should be looking at ways to use AI to reduce the desire for government.”

Desire is an odd word to use, because it implies an emotional need rather than logical need.

But for any system to exist and function effectively it requires it to be governed otherwise it tends to self destruction.

This is just as much true for humans as it is for physical machines and information processing systems.

In fact even apparently mindless colonies find benifit in acting as a system hence we see insects such as ants, termites, bees etc acting as collective systems. Even apex preditors such as wolves, hyenas, lions etc acting together in packs for mutual benifit.

At the very least there is safety in numbers as well as benifit by specialization. Even apparently issolated individuals benifit from a degree of colonization.

But other benifits are greater efficiency in energy utilisation thus production.

Humans have thus taken things to the point where we build force multipliers that work in concert to quickly achieve things that otherwise would not be possible.

None of these systems work without hierarchical control structures and a quite sophisticated level of mutual cooperation.

When such control structures fail to function correctly for the whole system, that’s when they start to break down and fall apart, and end in some way that is all to often catastrophic beyond the ability of the individual parts of the system to compensate for.

In effect this destructive end is what you appear to be arguing for…

Jamie Andrews • December 29, 2023 7:29 AM

Whilst the general idea that using some kind of IT to mediate democratic will is an old one that I have sympathy with, linking this to the current mania with LLMs seems unlikely to have a good result

All the problems with LLMs – that they have no transparency about how they generate output, that they are prone to hallucinations and that they will pick up lots of bias from the training text – are things that we do not want or need in our political decision process

On top of this all the current leading makers of LLMs are all neoliberals, in favour of concentrating power in their hands, believers in dictatorship as a great model of governance and generally not the sort of people you are looking for to design a fair and equitable system

Grima Squeakersen • December 29, 2023 8:20 AM

@Jamie Andrews: Exactly right. Lack of transparency is a particularly critical issue for auditing voting. Wait until there are several consecutive important ranked choice elections where the announced results appear to contradict the “will of the people”, as broadly interpreted after the fact. That would imo be a fail-proof recipe for producing widespread riot. How a government run on AI would handle such an event leads me to some frightening speculation.

Winter • December 29, 2023 9:04 AM

@Clive, Ron,

they start to break down and fall apart, and end in some way that is all to often catastrophic

Catastrophic indeed:
Hurricane Katrina
‘https://www.outsidethebeltway.com/katrina_federal_money_for_louisiana_went_to_pork_not_levees/

It turns out Louisiana has gotten more than its fair share of federal dollars for infrastructure but its own lawmakers thought the New Orleans levees were not a priority.

The Ca Camp fire destroying Paradise
‘https://www.latimes.com/environment/story/2023-11-08/its-been-5-years-since-californias-deadliest-wildfire-can-we-stop-it-from-happening-again

But evacuation failures were only one element of the devastation in Paradise, which was traced to a high-voltage PG&E transmission line that was nearly a hundred years old.

For instance, a new law will soon require ember-resistant defensible space zones within 5 feet of a home. Those rules would have made a significant difference in the Camp fire, she said.

Clive Robinson • December 29, 2023 10:13 AM

@ Winter, ALL,

Re : Misappropriation of resources by the corrupt.

“There is this eternal desire for lowering or eve abolishing taxes in the US.”

But why would people have a desire to get rid of a collective good?

There are many answers to this but the one underlying nearly all is,

“We pay but only they benifit, so why pay?”

The interesting thing is the “they” changes on who of the “We” you ask.

Which tells you some important things.

Firstly that the “they” the “we” blaim are almost always “invented” as an Orwellian foe to distract the “We” from the real benifit thieves and corruptors.

Secondly the misappropriation is so engrained it has actually become the system.

Thirdly when things are that endemic there will be an eco-system of dependents, who are the corruptors, and they will quite literally kill to keep their life style based on misappropriation and corruption.

Fourthly the stress this faux eco-system builds up as it grows, as more and more “jump on the gravy train”. Eventually it gets to the point it has to be forcefully exorcised from the body politic. This all to often leads to significant conflict as we have seen.

Fifthly as we are also currently seeing those who invent the Orwelian faux foe find they have in effect boxed themselves in, and the only way to maintain their position is to “create a hot war” and “profit from it greatly”. We’ve given it a series of cute names such as “Disaster Capitalism” and “War profiteering” over the years but as Shakespeare noted the odour changes not though the name does. The point is it’s a very real war with pointless death mass destruction and the transfer of even further masses of “treasure” into self selecting pockets.

This string of events happens with every political system where money becomes the major driver to success, and openness ceases under a carpet of secrecy called by the meaningless term “National Security” or similar. Which almost always hides the transfer of significant wealth from the public purse into the pockets of just one or two well hidden beneficiaries. Who all to often then use that wealth to buy the legislation and regulation they want that is all to often at significant variance to the wishes of those of the civil population that proportionately contributed most to that misappropriation of common good resources.

It does not take much thought to realise this is a self reinforcing degenerative spiral, the outcome of which can almost only ever be death, destruction, and enslavement… With those “We” of the commons being the ones who pay in every way on every step of the forced march to disaster.

emily’s post • December 29, 2023 11:13 AM

History has instances where the butler i.e. the major domo i.e. the mayor of the palace, in view of the exercise of practical power, overthrows and usurps the authority retained by the rightful ruler who delegated the practical matters.

The AI proposal here is certainly the ultimate instance of this delegation.

I, for one, welcome our new algorithmic overlords.

JonKnowsNothing • December 29, 2023 12:37 PM

@Winter, All

re: California ember-resistant defensible space

Fire maybe a universal human experience as a BBQ but fire containment is not universal.

In California

We have a number of jurisdictions that cover fire protection or fire containment. It is not one-size-fits all. There are other boundaries that apply too.

City, County, State, Federal and Combined Emergency Units all have responsibility to track or contain fires, if containment is possible.

So the important thing to know is Which One(s) Apply To You . If you do not know, as most MSM reporters do not, then you won’t know which one(s) are In Charge. And that is an important point.

To be sure, cities, counties, state, federal systems all have mutual support agreements in place for fire, police, disasters but there is a difference in response and responsibility.

Urban Fires

Most folks in the cities think that local city firefighters will respond to a fire in their home. But this is not always the case. Cities have un-incorporated areas where city rules do not apply. They are technically in the county, so the county firefighters are the ones to respond.

Rural Fires (1)

Most rural fires are under the county firefighters response area. There maybe a city fire station near you, but if you live over the city limit line, they may not be first responders.

State response is handled by CALFIRE. Rural homes are required to have 3,000 gallons of water dedicated solely for fire firing purposes. CALFIRE has extensive rules that affect rural homes and ranches but there are exceptions to those rules.

note: Not every home or rural location complies with Gov Regulations.
note: Land that does not have a Dwelling Unit (aka home) on it, aka bare land, does not need a CALFIRE tank.

Park Fires (2)

Forest fires are divided into State or Federal Parks. If it’s a State Park, then CALFIRE is responsible. If it’s a Federal Park it’s the National Park Service. If the fire gets large there is a combined incident cooperation command center activated.

As to the fire ember distance, this is a complicated area for “who is in charge” and in which county, city or fire district you are in.

In my area, rural defensible areas are 30ft-100ft depending on the structure or type of structure.

PG&E has a different fire requirement list and it often causes huge problems both in how it is implemented or not implemented.

PG&E has the right to clear cut anything under their major power lines ~50ft. That means TO THE DIRT. They go through orchards, forests, parks anything that is under the power lines can be cut down.

Some of the issues with this is that the path selected by PG&E for their power lines can go through ancient forests, existing farms, orchards or even cities and housing tracts. If it’s under the power lines they can all be removed. There have been a number of lawsuits and stop work orders given to the PG&E outsourced tree fellers when clear cutting hundreds year old trees because they would not take an alternate path to avoid the trees altogether.

The climate problem for PG&E is high heat and high winds. When the active wires clang together they spark and that’s a problem. Exploding distribution transformer on the utility poles also cause fires and sometimes environmental damage.

Underground is safer from wind sparking but not safer from flooding.

It depends on which disaster you want to avoid, and how much you are willing to spend.

In regards to PG&E, they are guaranteed a 3% profit by the CA Public Utilities Commission. It does not matter how much PG&E spends as those funds comes from the CA PUC, they will never have a dip in profits. It’s a Capped Profit arrangement.

===

1) State
ht tps://www.fire.ca.gov/

has an incident map for all active fires

2) Federal Combined

ht tps://inciweb.nwcg.gov/

InciWeb is an interagency all-risk incident information management system.
has an incident map for all active emergencies

Winter • December 29, 2023 12:43 PM

@Clive

“We pay but only they benifit, so why pay?”

In the US, the “they” were African Americans and later cumulative successive waves of immigrants.

It is the old Divide and Conquer.

The Netherlands had a quirk in this “They profit”. When everyone drowns if the levies fail, everyone will pay for good levies.

Hurricane Katrina flooding New Orleans showed that even that does not work in the US.

Anonymous • December 29, 2023 1:47 PM

Perfect communications channels do not lose any information.

JonKnowsNothing • December 29, 2023 4:20 PM

@Winter, All

This is a much better current reference…

Their land is sinking. But Tulare Lake farm barons defy calls to cut groundwater pumping (1)

Levies are big deals in California. Netherlands is not the only place under water. There are levies throughout California and we have our share of failures too.

Of curious note:

Central California was once an inland sea. Look at a satellite image from Baja California to Sacramento and you can see what was there a long time ago.

Much of this area is sand. Lots and lots of sand. Good for growing grapes.

Years ago there was a running joke (sometimes in practice) that people should buy desert property along the Arizona-California border because when the BIG ONE (earthquake) hits, most of California will fall into the ocean.

===

1)
ht tps://www. latimes. com/environment/story/2023-12-27/tulare-lake-land-barons-defy-calls-to-cut-groundwater-pumping

2)
ht tps://en.wikipedia.org/wiki/California_Delta

https://en.wikipedia.org/wiki/California_Delta#Levee_failures

The Delta was formerly located at the bottom of a large inland sea in the Central Valley, which formed as the uplift of the California Coast Ranges blocked off drainage from the Sierra Nevada to the Pacific. About 560,000 years ago, water breached the mountains, carving out the present-day Carquinez Strait and San Francisco Bay.

Ismar • December 29, 2023 10:39 PM

Democracy is too easy to hack and it is what happens today wherein minority of people act like parasites by taking advantage of other’s peoples shortcomings and unattainable desires. No AI or any other technology can fix what is essentially a problem of human nature .
In addition, why do we some people think that having an AI as a filter to knowledge is a good idea when all we get with AI suggested content is staying in the circle of knowledge we are already familiar with?

cls • December 29, 2023 11:58 PM

the AI cook is a really dumb idea. most of my cookbooks have an index, by ingredient. oh gosh, I have a bunch of spinach, not feeling creative, what can I make? yeah, retro concepts, like book, and index, and human ingenuity.

P/K • December 31, 2023 3:01 PM

Sounds nice, but probably the majority of people don’t have personal tastes and desires, they simply want to have what others have, do what other people do. Initially they looked at the traditional elites, nowadays at celebrities and influencers. Social media only increased that copy-cat behaviour, so for the time being that personalization sounds like a marketing story for the AI business.

Clive Robinson • December 31, 2023 6:14 PM

@ P/K, ALL,

“but probably the majority of people don’t have personal tastes and desires, they simply want to have what others have, do what other people do.”

In the UK we used to call that “Keeping up with the Jones”, and such people that behaved that way got scorned (but as you note times change).

In fact one UK sitcom had a woman who was an awful “social climber” who called herself “Bouquet” even though it was actually spelt “Bucket”.

https://en.m.wikipedia.org/wiki/Hyacinth_Bucket

The show was totally and utterly cringe worthy and I found it unwatchable due to it’s baaic premise of society as a game of snakes and ladders that was never won so…

But apparently the show was very very popular. In fact several people have analysed the character, and it’s been claimed you can tell your own social status from lower working class to upper middle class by how you see her. Apparently the more sympathetic you are the lower your status.

Not sure what my loathing of the show for it’s banality and predictability and the desire it induced to be somewhere else even “Doing the dishes” says about me but hey =^(

Which brings us to,

“Initially they looked at the traditional elites, nowadays at celebrities and influencers.”

I’ve a life long aversion for treating people by “rank or class” and it’s been remarked I take the “Prof Henry Higgins”[1] approach,

“The Prof claims his manners are the same as Colonel Pickering’s. To which Eliza disagrees, “That’s not true,” she says, “He treats a flower girl as if she was a duchess.” To which Higgins replies, “And I treat a duchess as if she was a flower girl.” Higgins continues, maintaining that good manners or bad manners are not important; instead, it is more important to have the same manners for all people.”

So “the same manners for all people” like it or lump it 0:)

As one of my then elderly relatives once put it,

“You gets what you get, take it or leave it, it’s all the same to me.”

[1] From George Bernard Shaw’s Pygmalion, made into the 1964 film “My Fair Lady”. Where Rex Harrison plays the Prof, and Eliza is played by Audrey Hepburn. One fun part is where Eliza’s father on hearing about the Prof taking Eliza into his house, goes around and trys to get money out of the Prof. It’s a speech on morals that every one can learn something from,

https://m.youtube.com/watch?v=zP43w5MCKqI

Clive Robinson • January 1, 2024 9:40 AM

@ Winter,

Re : Thus equality is still scaled.

“The difference between the Colonel and the Professor is that former values everyone and the latter despises everyone and let’s them know it.”

Not quite, the Colonel also “let’s them know it” as well.

The differece is not what they do, that is the same for both of them, it’s how they perceive others.

That falls on a scale with the colonel thinking of people in a courtly manner and the Prof not thinking of people as anything but animate objects.

It brings up the notion of how would you treat a Simulacrum with both intelligence and agency with respect to the natural version of what it is a simulation of (a subject that is going to arise with AI-Pets starting to appear).

So a real dog or a robot dog. For various reasons the robot will or should be limited in it’s agency such that it does not bite you etc and will be falsely friendly. With a real dog there is always the risk it will bite or even rip your throat out if it feals threatened etc.

If you’ve only ever interacted with Simulacrums then you would inherently believe they were “good”, “friendly”, or “courtly” and that would be a part of your core perception via cognative bias.

So if you then encounter a wild dog, your perceptions would be entirely wrong, and how you approach the wild dog would be wrong and your chance of getting your throat ripped out much increased.

So it could be argued that the Colonel “implicitly trusts” whilst the Prof “conditionaly trusts”.

Further that the Colonel puts himself at much greater risk, than the Prof does. Thus the Prof is more “street wise”.

Winter • January 1, 2024 9:58 AM

@Clive

So it could be argued that the Colonel “implicitly trusts” whilst the Prof “conditionaly trusts”.

No, the Colonel treats everyone like an equal, the Professor treats everyone like a lower being, an object. The Colonel treats Eliza like a valued woman. The Professor treats her like a tool.

That says nothing at all about “trust”. The Colonel is not necessarily easier to swindle.

Larry B • January 1, 2024 2:03 PM

With respect to AI affecting democracy this way: Wow…what a naïve and off-the-mark idea this is!!! First, even if such a system was allowed (which it would never be!), any social psychologist will tell you (and show you the data) that what people SAY they want and support is rarely even close to what they actually choose. So your supposed “interview” about what people want is flawed to begin with. An example in “real life” is the hopelessly inaccurate Nielson ratings from years back, when people (supposedly!) recorded what they were watching…which turned out to NOT be what they were watching! They “wanted” to be watching PBS, but they were watching sitcom junk.

But much more to the point, the notion is pathetically naïve that democracy even exists in our societies. It should be obvious to anyone that money and power controls the elections and the passage of laws. The “will of the people” rarely is allowed to prevail, and then only when there is overwhelming, and emotional, support for a single law.

Clive Robinson • January 1, 2024 5:55 PM

@ Winter,

Re : Equal to what being.

“No, the Colonel treats everyone like an equal, the Professor treats everyone like a lower being, an object.”

The Colonel does not treat people as equal to himself, nor does the Prof.

As I’ve already said,

“That falls on a scale with the colonel thinking of people in a courtly manner and the Prof not thinking of people as anything but animate objects.”

The Colonel simply puts other people at the top of a scale and as you say the Prof puts them at the bottom.

So they are both behaving in the same way, it’s just where the stick the pin in the scale…

We know very little about the Colonel from either the play or the musical.

But we do know that he’s set up not as an independent character but as a partial reflection of the Prof as both a plot and narative foil. In effect the pair are the “left and right shoulder ‘twin’ deamons” of an individuals personality.

If you look at the Colonel, whilst superficially different to the Prof in that he appears more genteel, he actuall has all the same failings as the Prof with regards Eliza and actually sees her as just “a horse in a race” on which he and the Prof “have had a small wager”. The Colonel behaves as the “owner” and the Prof as the “trainer”.

So in analysis the Prof comes off as the more honest of the “twins” he and the Colonel are in almost every way.

Watch the film through and you will see what I am saying is what Geroege Bernard Shaw was promoting as part of his “social commentary”.

If you want to see the characters written to be concerned and care for Eliza, her feelings, and well being first look the the Profs housekeeper then to his mother. With ultimately the Profs mother treating Eliza as her social and moral equal.

Winter • January 1, 2024 6:10 PM

@Clive

The Colonel behaves as the “owner” and the Prof as the “trainer”.

More like mother and father, respectively, Victorian style. The setup of the play (not the movie) is more like a family with understanding mother and callous father.

Professor Higgins was modeled after an existing phonetician, Daniel Jones. I do not know how he was in private life.

cheetour • January 2, 2024 5:10 AM

I disagree very heavily with the idea of implementing AI into the process of voting or lawmaking. There is no such thing as an apolitical, unbiased computer program; they are created and maintained by people, and usually, those people come from IT and finance, with no background in social policy, political science, or philosophy.

Mario Stipcevic • January 17, 2024 2:40 PM

Well,

With this kind of reasoning we won’t be needing to wait until AI esslaves us – if it even could. Because we are going to enslave us by it! If I understancd correclty, the democracy will work better if we let AI to chose politicians for us! What an idea!!

piglet • January 18, 2024 8:56 AM

This is probably the most ill-conceived piece I’ve ever read from Bruce Schneier. Many good criticisms in the comments. I hope Bruce will take this input seriously.

I’ll add just this: this looks like a typical example of a “solution” in search of a problem. The problem as Schneier paints it is an alleged lack of choices. Only dozens instead of thousands of choices at the restaurant! What a tragedy! This is bullshit. More consumer choice is perhaps the last thing our society needs.