Forum: Author Hangout

I play with AI for the same reasons I write: it's fun, I can learn something, and it gives me something to do. As for what it can produce as prose: superficial, pedestrian, and bland.

If I write a story, and use an AI, does the AI provide creative input? Can an AI be 'creative'?

If so, sure, give it an author credit, at least in a side note. But I don't think we've made that decision yet.

I will note as input to this discussion that the US Supreme Court has determined (just this past year) that AI can't be an author for purposes of copyright law. IIRC (but I'm not completely certain), they ruled that something AI generated isn't eligible for copyright protection.

For the moment, I'm virtually certain (at an I Am Not A Lawyer level) that a person could copyright anything an AI generated, though there could be some very interesting disputes over that if the AI-providing organization sued. There may also be license agreements explicitly allowing or denying copyrighting outputs.

There's also the monkey selfie case.
https://en.wikipedia.org/wiki/Monkey_selfie_copyright_dispute

In December 2014, the United States Copyright Office stated that works created by a non-human, such as a photograph taken by a monkey, are not copyrightable

That would cover AI as well.

During a hearing in January 2016, US District Judge William Orrick III said that the copyright law does not extend its protection to animals.[7][31] Orrick dismissed the case on 28 January, ruling that "if Congress and the president intended to take the extraordinary step of authorizing animals as well as people and legal entities to sue, they could, and should, have said so plainly.

Again, the reasoning here would cover AI as well.

PETA appealed, and the parties settled before the 9th Circuit issued a final ruling. The 9th refused to vacate the district court decision.

In April 2018, the Court of Appeals for the Ninth Circuit denied the motions to vacate the case.[42] On 23 April, the court issued its ruling in favor of Slater, finding that animals have no legal authority to hold copyright claims.

The tricky part is going to be the point at which we have AIs that pass every reasonable test for sentience we can throw at them. At that point, do the rules change?

Unlikely unless Congress decides to change the rules by changing the copyright act.

the most likely way we get to this point is a successful court challenge granting the AI personhood and the right to due process.

In my opinion, the only way that (a court granting AI personhood) could get any less likely would be for the probability of it happening to become negative.

On 23 April, the court issued its ruling in favor of Slater, finding that animals have no legal authority to hold copyright claims.

This decision doesn't exclude the human who arranged it, from claiming the copyright for the work for themselves.

As far as I understand, these are widely different legal issues (the work in question vs the person claiming copyright).

HM.

This decision doesn't exclude the human who arranged it, from claiming the copyright for the work for themselves.

Actually, yes it does. The final decision by the 9th circuit left the United States Copyright Office decision, that no copyright at all existed in the monkey selfie photos, standing.

How can author A's rights be protected if AI contributions and their origins are opaque?

They would have to sue author B exactly the same as if the same situation occurred with no AI involved.

And the legal situation would be no different if the AI contributions had been acknowledged by Author B. Author A would still have to sue Author B for the violation.

If author A finds something they wrote appearing verbatim in author B's work, presumably they have copyright protection (which may be impractical to enforce). If author B used AI and the AI contributed the material in question, presumably author B is still liable for the breach of copyright.

Take it one step further. Your AI and my AI independently produce verbatim texts however the instructions we gave our AI's were NOT identical. Who has copyright?

A similar issue occurs with AI driven vehicles, when an accident occurs. Who do you sue when the AI vehicle cripples you? Who gets punished? Would it be just and fair for a human driver to be imprisoned if at fault but an AI owner to just pay a fine? What happens when two AI vehicles collide and the passengers are killed?

Who do you sue when the AI vehicle cripples you? Who gets punished? Would it be just and fair for a human driver to be imprisoned if at fault but an AI owner to just pay a fine?

It would be an super rare case for a non-intoxicated human driver to be criminals convicted for an automobile accident.

As for coverage for damages, one possibility is a set, no-fault indemnification scheme, something like Worker's Comp. Injured in an accident involving autonomously-driven vehicle? Manufacturer pays amount determined by severity of injury, plus expenses without any investigation of causality or negligence.

It would be an super rare case for a non-intoxicated human driver to be criminals convicted for an automobile accident.

If you call over a hundred a year i the UK alone "super rare", sure.

without any investigation of causality or negligence

Excellent way to ensure that common causes are ever identified or corrected.

If you call over a hundred a year i the UK alone "super rare", sure.

Really? Wow. I'd love to know the details. I was thinking of the US. Here, in most states, vehicular homicide, even when drunk, wouldn't get more than a few years. Now, intentionally using a vehicle as a weapon would be different, but then that wouldn't be possible with an autonomous vehicle.

Excellent way to ensure that common causes are ever identified or corrected.

Hasn't seemed to work that way for workers comp. Big incentives for the company to eliminate losses, both financially and from a PR perspective. Also nothing to prevent government involvement with standards, inspections, etc.

Edited for stupid mistakes!

but then that wouldn't be possible with an autonomous vehicle.

Unlikely isn't the same as impossible. Also a tech who maintains the vehicles could either screw up, or take the bribe and engineer an "accident" to a given target… perhaps unlikely but NOT impossible.

Eager for another blow job huh?

;)

Now, if a person had Chat GPT create a story, then the human merely edited it. I think that should be declared.

Someone has. And, looks like a LOT of work to keep the current models following the plot and the result is not very good:

https://www.reddit.com/r/ChatGPT/comments/120oq1x/i_asked_gpt4_to_write_a_book_the_result_echoes_of/

I'm gonna be honest. I didn't play around with AI as much as others, because I know that AI can't give you anything original. But, when I did play around with it, the responses I got to my writing prompts were AT MOST 500 words long.

So, unless you all are using some AI that is capable of writing actual stories, I don't see that as anything more than getting a premise to work with.

If you're like me, you'll look at the result, then get inspired to modify at least some. In fact, out of the ten names Chatp-GPT4 generated, I took one! I did let him stay a 6'1" forward, but as a freshman, not a senior. Part of the reason I didn't take more was that the names were so boring. But, as I said, just seeing the initial work freed my brain up to do what I wanted.

But, as I said, just seeing the initial work freed my brain up to do what I wanted.

So does LSD or so I heard.

Cool. Scrivener also has a name generator, and it can be set for particular nationalities. I've occasionally found it helpful when trying to get a name that "felt" right.

Thought experiment: All vehicles are replaced overnight with self-driving cars at the current state of the art. What happens to the rate of death and injury from auto accidents?

(My bold)

If you are lucky the numbers don't noticeably change.

Most likely result is a sharp increase followed by massive public backlash and driverless cars are outlawed on public roads or in public spaces.

Current tech simply isn't capable.

Human may be vastly superior when not intoxicated, distracted, or asleep. The tradeoff would be those against AI error. And if all cats were AI driven, the need to counter static actions by other drivers would drop.

. And if all cats were AI driven

I've never heard of AI driven felines before.

Well, they're crazy enough as they are now. No need to add AI weirdness to their ingrained eccentricities!

Thought experiment: All vehicles are replaced overnight with self-driving cars at the current state of the art.

@joyR:

Current tech simply isn't capable.

@awnlee jawking:

Humans are vastly superior because of their ability to draw conclusions from incomplete data. for example, a human in full control of their faculties would never assume a white lorry was the sky

Both your comments implicitly assume a mixture of human driven cars and autonomous cars without central support nodes and without assisting systems in the roads.

Current tech is capable of doing this. Its the mixture with humans driven old systems which makes it so complicated.
If the white lorry has the new technology it would be reporting its position and other driving data like speed, cross weight, ... to the next central node, which itself communicates this data to all AI-driven vehicles in the vicinity, making it unnecessary to "assume" anything about the white lorry. Your car's AI can then directly negotiate with the lorry's AI how to best pass it.

The real big problem with such a solution:
It can't be done overnight. It had to be done in small areas at a time and then expanded to finally cover the whole country. During this transition time manual driving within an autonomous driving (AD) area must be made technically impossible, as autonomous driving outside of AD-areas.
If you use a dual-mode car or a single mode car (either autonomous or manual driven) then depends on your needs and preferences.

Is it politically possible? I doubt it.

HM.

The real big problem with such a solution:
It can't be done overnight. It had to be done in small areas at a time and then expanded to finally cover the whole country. During this transition time manual driving within an autonomous driving (AD) area must be made technically impossible, as autonomous driving outside of AD-areas.
If you use a dual-mode car or a single mode car (either autonomous or manual driven) then depends on your needs and preferences.

There's another problem that's probably just as big with this idea. Replacement of the entire fleet of older manual driving only cars. The cost of this would probably be as much as the cost of the infrastructure for your proposed centrally managed autonomous driving.

During this transition time manual driving within an autonomous driving (AD) area must be made technically impossible, as autonomous driving outside of AD-areas.

Replacement of the entire fleet of older manual driving only cars.

I guess this transition time is at least one decade. Most cars are replaced by then, hopefully by dual mode cars or AD-only cars.

BTW, the infrastructural cost wouldn't be such high because of the far lower cost of the car's autonomous driving unit. (less sophisticated, they don't have to deal with human drivers and their flaws).

HM.

Most cars are replaced by then

Most but not all. There are Model T cars that are still drivable.

hopefully by dual mode cars or AD-only cars.

I wouldn't recommend betting on public acceptance being anywhere near that fast.

I guess this transition time is at least one decade. Most cars are replaced by then, hopefully by dual mode cars or AD-only cars.

A person may well replace their car with a new model every ten years or less, but the car does not evaporate. It is either traded in and will be resold or sold at auction, privately or in many cases passed down to a son or daughter.

Currently in the UK one in four cars are over thirteen years old (About 8.4 million)

In the US in 2020 one in four vehicles were at least sixteen years old.

The real big problem with such a solution:

You missed out several very big problems:

An AD (to use your nomenclature) will slow and stop if an object is detected in front of it that cannot be avoided. City streets full of cars and anyone who steps out into the street knows the AD cars will stop. Given human nature those cars will be constantly stopping for people, cyclists, motorcycles, disability's scooters etc etc. Not to mention the kids delighting in causing traffic to grind to a halt.

Of course in your scenario no vehicles from other countries could be allowed on the roads, nor agricultural vehicles.

Are you sure current tech can cope with heavy traffic in adverse conditions?

Then there is the current push to force EV vehicles. When multiple vehicles stuck in traffic run out of battery power they would no longer be connected to your 'system', not to mention many are not designed to be towed and require lifting or charging in place.

So no, I don't think it is possible with current tech and even less possible in Europe where vehicles are constantly moving between countries.

The system to detect more than one animal is prohibitively expensive.
Basically it fails when it's two deer.