In general, all areas where truth and fact are subjective enough to never really matter. An unprosecutable liar is the perfect scapegoat in future litigation.
i didn’t belive it was possible to be getting more cynical in my dotage…
1 Like
I think you’re probably being overly cynical. Just as a driver is responsible for the behavior of the vehicle, the person who commissions work from an AI will be responsible for its output.
(With, as with vehicle liability, arguments about whether manufacturer defects were to blame.)
I think this will become the prevailing opinion. The WGA maintains that AI can’t author anything by itself any more than Scrivener can. They say the person prompting the AI is an author, and in Hollywood, will need to belong to the union.
There’s also a lawsuit against the US Copyright Office to reverse their decision that AI-generated art is not protected by copyright, on the basis that the crafting of prompts is an act of authorship.
I think there are powerful market forces that want AI-generated material to be protectable IP, and also to hold AI-generated art to the same legal standards as all other art. It’s just going to take a little while to get sorted.
1 Like
The law will need to be tested in the courts on a jurisdiction-by-jurisdiction basis, which may take years or even decades to fully resolve.
The two basic analogous precedents that are obvious references cases that could form the basis of future cases are:
- vicarious liability of employers for the tortious actions of their employees (where the human “employer” could be deemed liable for the negligence of the AI “employee”), and
- the wide liability net in defamation actions for any party deemed to be publishing or republishing the offending material, which could hold numerous parties in the production chain responsible, from the company that made the AI alogrithm, the person who wrote the source material used to train the model, the person using the AI model to produce the offending output, and anyone who repeats it (including in theory the host of the internet forum someone innocently posts a snippet on).
In both these examples, neither a lack of knowledge of the failure nor the absence of any ill intent would save a party from personal liability. It’ll be fun to see how this pans out.
The copyright principles that apply here have been set in the famous “monkey selfie” case, which is well worth a read if you’re interested in the subject and aren’t familiar with the details. The TL;DR output f it is that a monkey can’t be an author (much like an AI can’t) because it’s not a person. Therefore the product wasn’t authored at all and can’t be subject to copyright protection. So the guy that set up the camera, and handed it to the monkey to press the button, wasn’t able to claim the copyright either. It would take a deliberate intervention of a higher court or legislation to reverse this basic principle… ie a lot of powerful lobbying.
I don’t think that’s clear at all.
I’m familiar with the case; I don’t agree with your analysis.
The argument being made now in federal court is that the AI is no more the author than Scrivener is the author of your book. The only difference is the power of the technology and the sophistication of the interface used by the author (a person) to direct the technology.
A clearer comparison is the use of “content aware fill” in Photoshop to remove people from a photograph that will result in a digital image that is protected under copyright. The author uses a pointing tool (a mouse or tablet) to indicate to the computer where to use AI to paint new pixels into the image. The author does no painting themselves, and the AI makes creative decisions about how to paint the new pixels. The copyright office accepts this as an authored work.
The AI artist suing them argues that the prompt he wrote is simply a different kind of interface, and the AI a simply more powerful version of the “content aware fill” that already qualifies for protection. The prompt is an act of authorship, and it is the human using the interface that is the author, not the computer software following the instructions from the interface. In this case, the prompt needed to direct the AI to paint that particular painting is very long, complex and specific. It will be difficult for the Copyright office to make the argument that in the face of that complex prompt that the AI was acting on its own agency like the monkey.
Only time will tell.
1 Like
No, the position in that case is that using a AI tool on a specific area does not invalidate the pre-existing copyright that exists in the underlying artistic work.
If I were on your side of the argument, I’d be trying to make the case that, although the AI generated materials don’t have protection, sufficient creative effort is being overlaid on the raw materials after the fact by a human to create a new artistic work capable of protection.
Even then, though, it’s analogous under current law to taking a photo. The AI is the beautiful landscape. At best you can stop me using your exact photo of that landscape, but you can’t stop me pointing my own camera at it. Ie, the next guy that uses the same AI programme and gets a very similar output won’t and can’t be copying you.
This is (partially) why map makers are allowed to have a certain number of deliberate mistakes in maps per set area. It’s ironically the copying of the errors that is protected*+*.
(+ - gross oversimplification)
That’s why watching it unfold will be fun! The biggest lobbying budget will likely decide the outcome.
1 Like
I only hope that the traditional slow pace of the legal system will be able to keep pace with the onslaught of what AI will bring to not only the marketplace, but to our very democracy. Although we’ve had a few hundred years to withstand the onslaught of the reality TV star becoming president, for the first time in decades, I woke up this morning, not in a cold sweat, but in a mini-panic at the prospect of having the iterative power of Almost Intelligence unleashed for those gerrymanderers who wish to draw the most demonic electoral maps to ensure their hold on their states legislatures, and ultimately, the electoral college that is in reality, the manufactured electorate that decides the presidency.
I’ve posted about how the AI timeline for the control of society, government, war, finance, medicine, religion, and countless other aspects of life on this planet, will mimic the century-and-a-half evolution of fossilized carbon as the dominant factor in our development as a species, but at an order-of-magnitude faster rate that the relaxed, century-and-a-half evolution we experienced with fossilized carbon.
Now, I’m convinced we may be looking at a rate of the development (more likely intrusion) of Almost Intelligence into everyday life that may be multiple orders-of-magnitude faster than the century-and-a-half we had with fossilized carbon. Not good. When those who are the most familiar with Almost Intelligence talk about how it will change everything, I am beginning to have a glimpse of just what they mean.
The redrawing of election maps to benefit one political party or another is just the most recent mini-panic I’ve had regarding AI. Tomorrow, and the days that follow, I would not be surprised to wake up to any number of mini-panics regarding issues that we are NOT ready for. The only limit appears to be my demonic imagination at what others have already researched and planned as potential game changers for us all.
I keep reminding myself of the notable gains made with the 200 million proteins who’s structures were ALL solved by AI … an accomplishment that boggles the mind, particularly in the danger it represents if and when it is made available for nefarious purposes.
All this represents a major challenge to those of us who can use our imagination to write about the challenges that we face. Government cannot begin to address the issues that have not been recognized, placing a newfound critical importance on those who put fingers to the keyboards or speech-to-text apps and render those issues into reality, hopefully before the electoral maps have been re-gerrymandered to hollow out our democracy.
scrive
Don’t worry, it’s going to be way worse than a bit of election rigging (wasn’t that a conspiracy theory?).
2 Likes
Your panic is several years too late. Gerrymandering a map is a fairly simple “optimization” problem that computers are already more than capable of solving. Gerrymandering is restrained (if it is) by the legislatures, the courts, and ultimately the voters.
1 Like
Given that it’s not yet really started to get a handle on the internet and social media, let alone keep pace, I wouldn’t put much faith in it.
1 Like
That’s what they want you to think. 
2 Likes
It may be …
It is a stretch for me to understand the challenges that existed with ALL the possible permutations involved with discovery of how 200 million proteins fold.
With the very limited scope of understanding that I can draw on, determining how proteins are folded presented perhaps a perfect challenge for Almost Intelligence to solve.
(This is where I am on very thin ice).
I imagine there are very specific gradations of rules that the carbon and other atoms are known to have to follow when folding, or at least are most likely to follow, for there to be a successful folded protein. (Here is perfect opportunity for experts in biology to enter the discussion)
So the power of Almost Intelligence to run the countless permutations that are needed to determine the precise configuration to have a successful protein structure is the exact threat that I see exists for a new wave of gerrymandering that can escape even the most careful scrutiny by the legislatures, the courts, and ultimately the voters.
Yes, eventually, those entities will eventually realize what is happening with those jurisdictions that have been hacked, despite the popular vote.
Where I am coming from is a place where even a few decades, and possibly even a few years could make the difference in what countless generations will have to deal with going forward, given what most of us have come to realize of the world around us (and have yet so much more to learn) about the legacy that we are creating in realtime.
If the challenge we face represents a:
then the legislatures, the courts, and ultimately the voters will suffice.
The challenge, however, I believe has become a real time issue for which we may not be prepared, similar to how fossilized carbon changed everything, but now at a rate that may be orders of magnitude faster than humanity has ever experienced.
Think North Korea and climate change, together, on steroids.
AI may represent a test for humanity like none we have never faced, one that we unwittingly have brought upon ourselves, but cannot, and will not, undo.
That is why, as I mentioned earlier, those of us who have both the desire and skills to put the words, in whatever form, coupled with our collective imaginations, need to peer into what may lie ahead, to open the literary spigots of thought to begin the task of preparing us for what may lie ahead.
We have a worldwide inventory of enough bullets, bombs, aircraft, submarines, soldiers, gunpowder, TNT, fossilized carbon extraction and other unmentionables to do unimaginable, existential harm to humanity.
What we need are the imaginations and communications necessary to bring the power of AI to the fore so that the legislatures, the courts, and ultimately the voters can understand what some of possible outcomes from AI, representing our own supercharged aspirations, may be.
Can we somehow match the power of AI with our imagination of the possible outcomes that such AI derived aspirations represent?
Buckle up!
scrive
How do you prepare for the end of your species? I’ve never done that before.
2 Likes
To design a political map, you need to know how many people there are and where they live. You don’t know how they vote, but you can guess based on party registration, demographics, and so on. But you can’t move the people, you can only draw lines in two dimensions, and each district needs to be within a given range of the ideal population. That’s why the optimization problem is relatively simple.
There’s also a point of diminishing returns. You can give yourself “safe” districts, in which your representatives are extremely unlikely to lose. And you can give yourself a lot of districts, in which your representatives can win a majority in more than their population-share of districts, but it’s very hard to do both. The more districts you have, the more marginal each of them will be. That’s why AI doesn’t help you (enough): you can optimize for safety or for number of districts, but not both at once.
And finally, in the US, you can’t redraw the state lines. Which means you can’t gerrymander the Senate, the Presidency, or the state Governorships, and you can only gerrymander Congressional districts on a per-state basis. So your AI doesn’t have as many degrees of freedom to play with.
1 Like
Finishing my thought … protein folding is hard because there are lots of degrees of freedom. The angle of bond A doesn’t really limit the angle of bond B, and the whole structure is three-dimensional.
1 Like
This is not humanity’s first rodeo show. Humanity has been here before, but we didn’t give up.
Over a half-century ago, we were looking down the barrel at nuclear annihilation … the 1959 movie “On the beach” was perhaps the first instance I can remember that told the story of our annihilation, followed a few years later by the hyper-real Cuban missile crisis.
1959 was pre-Vietnam War, pre both the Soviet and U.S. versions of Afghanistan, pre the fall of the Soviet Union, pre Iraq war, pre 9/11, pre internet, and certainly pre the bulk of virtually everything we know and have today that we don’t consider dated.
And yet the world worked with what they had … corny and impossibly dated by today’s standards … yet timely for the time. But it started people talking.
In many ways it laid the groundwork for a topic as surreal as climate change and AI is for many people today, and just as controversial as the most controversial topics of our day.
Start with 1959 if you want to get an idea of how the generation before us stumbled, and fumbled, but kept trying, for there were hundreds, possibly thousands of followup references that were spawned by the movie.
Today, we have an embarrassment of riches, and array of tools we can draw on to tell the story … hell Scrivener could not have even have been imagined back then, and yet humanity somehow is still here … to address the multiple existential threats that now we face.
This is not an answer; it is possibly a start to remind us of the tools we are blessed to have - sometimes at out fingertips, and all around us.
By writing, and communicating, we begin.
scrive
P.S.Watching the movie today, considering the mood at the time, I am amazed that it was ever made, and how people watched.
1 Like
I know the movie (and the version from 2000) and I get your point. Yes, in general it works out just fine, despite the worst fears in face of <insert favorite doom and gloom scenario here> at that time. It really does.
This time we’re about to create a problem that is smarter than us. On purpose. That’s a first. And likely a last. Most people can’t even imagine what that means. Spontaneously visualizing a five-dimensional space in your head is way easier. (Don’t look it up.)
As I’ve written before, it doesn’t necessarily culminate in a Terminator-like “man vs. machine” scenario. Actually, that’s the least likely outcome. But are we, for instance, prepared to (at least try to) enslave our sentient and conscious creation?
Most likely people (who can afford it) will be tempted to enhance themselves in order to stand a chance in this new arms race against AI and other enhanced people (who can afford it). Do you think you will be one of them? Even so, who or what will you be at this point?
1 Like
That’s a different ending than the one I watched a day ago …
Closer to the story line of the 1966 science fiction novel Colossus by Dennis Feltham Jones, that was made into a 1970 American science fiction thriller film from Universal Pictures, titled Colossus: The Forbin Project?
(See Colossus: The Forbin Project - Wikipedia and Colossus: The Forbin Project (1970) - IMDb. Unfortunately, I was not able to find a free online rendering of the film.)
The story line posits a sentient super computer that is given absolute power over the U.S. nuclear arsenal. After communicating with it’s Soviet counterpart, the sentient super computer exerts complete control over all of humanity.
Turns out, reflecting on our possible state of affairs over a half-century later with how AI could evolve in the not too distant future, the Colossus book and movie were about as prescient as anyone can ever imagine. From 1966.
Before the vast majority of the assets of reality that we live with, and take for granted, today. Pretty amazing.
That is one direction we may take … we’ve already ‘enslaved’ all non-sentient computers to do our bidding … will that change once they are sentient and conscious. I do not know …
Or, can we co-exist … Colossus posits that we cannot … maybe suggesting that until such consciousness can mature toward accepting all sentient beings for what they offer, it may be a rough road as the AI’s evolve.
One might ask: Is Colossus merely a reflection of humanity’s own immaturity toward other sentient beings (including our very own)?
scrive
1 Like
Yeah, but you watched a movie, not reality. Wasn’t that your whole point of mentioning it?
Well, right now they don’t give, excuse my French, a fuck. That will change then. Morally, legally, everything.
Most people don’t hate ants (well, maybe they do in their homes, but in general). They are fascinated by them. And then they step on them by just walking from A to B.
Definitely yes. That’s part of the problem. At least part of that part of the problem where we have a say in it (see also: ants).
1 Like
1 Like