Given that current IP law requires the inventors to be human, who owns
the IP of inventions created by AI?
If the owners of a company that made an invention through AI claim it is theirs, and someone challenges their ownership based on the allegation
that the invention came from AI, would the patent be defensible?
For example there are companies out there that openly claim to be using
AI to accelerate their business development processes. So it may not be difficult for people who want to challenge AI-generated patents to prove
that the patents were in fact not derived from a human.
https://www.youtube.com/watch?v=IrLJEzUxIs0
Given that current IP law requires the inventors to be human, who owns
the IP of inventions created by AI?
We are getting quite close to the singularity where AI becomes more intelligent than the smartest humans. For some narrow problem domains
chess and go the machines already have overtaken mere humans.
ISTR someone tried and failed to get an AI registered two years ago as
the inventor of a patent in the UK supreme court:
https://www.reuters.com/technology/ai-cannot-be-patent-inventor-uk-supreme-court-rules-landmark-case-2023-12-20/
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns
the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to
produce its output.
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns
the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to
produce its output.
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns
the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to
produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a 'substantial part' of one or more original copyright works. As opposed
to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated
the AI is.
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns
the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to produce its output.
Hardly. There is no copyright in information, only in the way it is particularly expressed.
On 31/12/2024 7:14, Jon Ribbens wrote:
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns >>>>>> the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to >>>>> produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As opposed
to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated >>> the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
How do you determine that the AI has unlawfully copied works to create a derivative?
And how is this different from many human inventions?
Would you say that all the patents granted to human inventors up to now
were not "unlawfully" copied or derived from someone else's works?
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns >>>>> the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to
produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As opposed
to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated
the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
On Mon, 30 Dec 2024 11:54:17 +0000, Martin Brown wrote:
We are getting quite close to the singularity where AI becomes more
intelligent than the smartest humans. For some narrow problem domains
chess and go the machines already have overtaken mere humans.
Not really.
When "AI" can strip out all the crud from a webpage, then maybe - *maybe*
- I will reconsider.
The list of what "AI" can't do is still larger than the list of what it
can do. And even then that's in a very curated environment.
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns >>>>> the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to
produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As opposed
to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated
the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
On 31/12/2024 09:31, Spike wrote:
I’m still waiting for a list of things that AI has discovered that weren’t
known before.
The earliest one was probably the verification of the proof of the 4
colour map theorem (guided by humans as most such tools still need to
be). That was back in 2005 using a general purpose theorem prover.
Jon Ribbens <jon+usenet@unequivocal.eu> wrote:
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns >>>>>> the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to >>>>> produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As opposed
to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated >>> the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
Quite.
I’m still waiting for a list of things that AI has discovered that weren’t
known before.
On 31/12/2024 09:31, Spike wrote:
Jon Ribbens <jon+usenet@unequivocal.eu> wrote:
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns >>>>>>> the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to >>>>>> produce its output.
Hardly. There is no copyright in information, only in the way it is >>>>> particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As opposed >>>> to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated >>>> the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
Quite.
I’m still waiting for a list of things that AI has discovered that weren’t
known before.
The earliest one was probably the verification of the proof of the 4
colour map theorem (guided by humans as most such tools still need to
be). That was back in 2005 using a general purpose theorem prover.
https://en.wikipedia.org/wiki/Four_color_theorem
Earlier AI symbolic algebra tools found some cute results in general relativity too but they were credited to the author(s) of the code.
More recently Google DeepMind is now as good as a silver medalist at the International Maths Olympiad (and that is *seriously* good). I grant you
that isn't general intelligence but it is now very close to being able
to beat all but the very best mathematicians on the planet.
https://ukmt.org.uk/deepminds-ai-achieves-breakthrough-in-solving-international-mathematical-olympiad-problems
and at mathematics research level AI tools are getting very good now
https://www.scientificamerican.com/article/ai-will-become-mathematicians-co-pilot/
AI already has the edge over humans for large scale chip design.
On 31/12/2024 03:14, Jon Ribbens wrote:
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
That is debatable for the AI derivatives of Google's Alpha-Go-Zero.
It was able to boot strap itself up from the rules of Go to beyond human levels of play in about 2 months. In the process it found novel live
puzzle positions that had never been recorded in over two millennia of
human play. That is by any reasonable definition non obvious creative invention even if it was largely done by a combination of brute force
and very sophisticated Monte-Carlo search algorithms.
AI can also learn to imitate a human voice accurately from a worryingly
small sample of the spoken word. Telephone banking with "My word is my password" will cause a lot of trouble in the not so distant future. R4
You and Yours tested it on air fairly recently and got in first time.
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns >> >> the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to
produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a 'substantial part' of one or more original copyright works. As opposed
to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of copyright works, but there might be of some, depending how sophisticated the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
Martin Brown <'''newspam'''@nonad.co.uk> wrote:
[quoted text muted]
[quoted text muted]
That says that the Four Colour Theorem was proved in the 1990s,
presumably before the coming of AI.
On 2024-12-31, Martin Brown <'''newspam'''@nonad.co.uk> wrote:
[quoted text muted]
You've just explained why it was neither creative nor invention. This
sort of thing is a minor evolution of algorithms which were explained in hobbyist computer magazines in the 1980s. The main difference is that computers have become spectacularly fast with enormous amounts of
storage.
On Tue, 31 Dec 2024 03:14:44 -0000 (UTC), Jon Ribbens wrote...
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human,
who owns the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to >> >> > produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As opposed
to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated >> > the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
Merely being derivative isn't the test.
Earlier in the thread, someone linked a Kipling poem. Ignore for the
sake of argument the fact that Kipling's copyright has expired.
If I were to summarise that poem in my own words, it would be
derivative. But since it would be my words, not Kipling's, it wouldn't infringe his copyright.
On the other hand, if I were to write a short story which quoted a substantial part of Kipling's poem, that would infringe his copyright
(unless one of the Copyright Act exceptions applied). Typically in such circumstances, my publisher would seek permission.
On Tue, 31 Dec 2024 03:14:44 -0000 (UTC), Jon Ribbens wrote...
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who
owns the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order
to produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As
opposed to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how
sophisticated the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
Merely being derivative isn't the test.
Earlier in the thread, someone linked a Kipling poem. Ignore for the
sake of argument the fact that Kipling's copyright has expired.
If I were to summarise that poem in my own words, it would be
derivative. But since it would be my words, not Kipling's, it wouldn't infringe his copyright.
On the other hand, if I were to write a short story which quoted a substantial part of Kipling's poem, that would infringe his copyright
(unless one of the Copyright Act exceptions applied). Typically in such circumstances, my publisher would seek permission.
On Tue, 31 Dec 2024 13:22:01 +0000, Jon Ribbens wrote:
On 2024-12-31, Martin Brown <'''newspam'''@nonad.co.uk> wrote:
[quoted text muted]
You've just explained why it was neither creative nor invention. This
sort of thing is a minor evolution of algorithms which were explained in
hobbyist computer magazines in the 1980s. The main difference is that
computers have become spectacularly fast with enormous amounts of
storage.
*All* "AI" is just sophisticated pattern matching on steroids.
It may eventually be determined that is all "intelligence" is anyway.
However no one has even come close.
The problem with all this bollocks about "AI" is that we still can't
_really_ define the "intelligence" bit. "Artificial" is easy.
On 31/12/2024 17:07, Jethro_uk wrote:
On Tue, 31 Dec 2024 13:22:01 +0000, Jon Ribbens wrote:
On 2024-12-31, Martin Brown <'''newspam'''@nonad.co.uk> wrote:
[quoted text muted]
You've just explained why it was neither creative nor invention. This
sort of thing is a minor evolution of algorithms which were explained
in hobbyist computer magazines in the 1980s. The main difference is
that computers have become spectacularly fast with enormous amounts of
storage.
You are probably right to be afraid of it. Alpha Go Zero is in a
completely different league to its ancestors like DeepBlue (Chess) or
even more distant Samuels Chequers(Draughts) program.
I never expected to see a machine that could play Go in my lifetime. The
move tree explodes way too quickly with search depth on a 19x19 board.
Today it is hard to find any commercial chess program that doesn't play better than DeepBlue (dedicated hardware) did back in 1997.
*All* "AI" is just sophisticated pattern matching on steroids.
It may eventually be determined that is all "intelligence" is anyway.
However no one has even come close.
It is fairly likely that "intelligence" *is* just sophisticated pattern matching on steroids. Most human inventiveness comes from seeing
connections between disparate fields of endeavour that can be exploited
to obtain a novel solution to a long standing problem.
It is particularly true when a new development in mathematics suddenly
allows intractable physics problems to be represented in novel notation.
The problem with all this bollocks about "AI" is that we still can't
_really_ define the "intelligence" bit. "Artificial" is easy.
I suspect that we will know it when we see it.
AI's that are as smart as people (and probably smarter than most people)
are not that far away now. One more order of magnitude improvement in computational performance ought to do it. Then the problem will be doing
it using a lot less power than is required at present.
My own view is that intelligence and possibly self awareness are
emergent properties of any sufficiently complicated computational
network.
My own view is that intelligence and possibly self awareness are
emergent properties of any sufficiently complicated computational network.
On 31/12/2024 17:46, Martin Brown wrote:
My own view is that intelligence and possibly self awareness are
emergent properties of any sufficiently complicated computational network.
'The moon is a harsh mistress' was a good read back in the day. Sad
ending though IMO. Heinlein.
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Tue, 31 Dec 2024 03:14:44 -0000 (UTC), Jon Ribbens wrote...
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human,
who owns the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to >> >> > produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As opposed >> > to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated >> > the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
Merely being derivative isn't the test.
Earlier in the thread, someone linked a Kipling poem. Ignore for the
sake of argument the fact that Kipling's copyright has expired.
If I were to summarise that poem in my own words, it would be
derivative. But since it would be my words, not Kipling's, it wouldn't infringe his copyright.
I don't think that's right. For example, there have been plenty of
Sherlock Holmes TV programmes and such, because the early Sherlock
Holmes stories were out of copyright, but until recently none of them
tended to mention him keeping bees, because that idea was introduced in
the last story, which was not out of copyright. Even though none of the programmes wanted to literally copy the words from the story, they
couldn't copy the ideas either.
On the other hand, if I were to write a short story which quoted a substantial part of Kipling's poem, that would infringe his copyright (unless one of the Copyright Act exceptions applied). Typically in such circumstances, my publisher would seek permission.
"AI" does *nothing but* quote other peoples' works, mixed together.
On Tue, 31 Dec 2024 03:14:44 -0000 (UTC), Jon Ribbens wrote...
On 2024-12-31, Tim Jackson <news@timjackson.invalid> wrote:
On Mon, 30 Dec 2024 13:48:50 +0000, Norman Wells wrote...
On 30/12/2024 12:21, Jon Ribbens wrote:
On 2024-12-30, J Newman <jenniferkatenewman@gmail.com> wrote:
Given that current IP law requires the inventors to be human, who owns >>>>>> the IP of inventions created by AI?
The millions of people whose copyright the "AI" infringed in order to >>>>> produce its output.
Hardly. There is no copyright in information, only in the way it is
particularly expressed.
That idea/expression dichotomy rather depends on the individual
circumstances of each case.
The test is whether, in any particular case, the AI has reproduced a
'substantial part' of one or more original copyright works. As opposed
to just taking the underlying ideas.
I agree that there is unlikely to be infringement of "millions" of
copyright works, but there might be of some, depending how sophisticated >>> the AI is.
The "AI" isn't capable of dealing with ideas. It just mixes together
the works it has unlawfully copied and produces something derivative.
Merely being derivative isn't the test.
Earlier in the thread, someone linked a Kipling poem. Ignore for the
sake of argument the fact that Kipling's copyright has expired.
If I were to summarise that poem in my own words, it would be
derivative. But since it would be my words, not Kipling's, it wouldn't infringe his copyright.
On the other hand, if I were to write a short story which quoted a substantial part of Kipling's poem, that would infringe his copyright
(unless one of the Copyright Act exceptions applied). Typically in such circumstances, my publisher would seek permission.
Earlier in the thread, someone linked a Kipling poem. Ignore for the
sake of argument the fact that Kipling's copyright has expired.
If I were to summarise that poem in my own words, it would be
derivative. But since it would be my words, not Kipling's, it wouldn't infringe his copyright.
On the other hand, if I were to write a short story which quoted a substantial part of Kipling's poem, that would infringe his copyright (unless one of the Copyright Act exceptions applied). Typically in such circumstances, my publisher would seek permission.
Although what publishers do, perhaps out of excessive caution, isn't
really much of an indication. An example has been given here of
Sherlock Holmes keeping bees being a no-no for publishers to repeat.
But I doubt very much whether just that, without quoting a considerable amount of Conan-Doyle's actual words can amount to copyright infringement.
It can be quite amusing to ask Siri, Alexa, et al to open the pod bay doors.
On Tue, 31 Dec 2024 18:10:04 +0000, Norman Wells wrote...
Earlier in the thread, someone linked a Kipling poem. Ignore for the
sake of argument the fact that Kipling's copyright has expired.
If I were to summarise that poem in my own words, it would be
derivative. But since it would be my words, not Kipling's, it wouldn't
infringe his copyright.
On the other hand, if I were to write a short story which quoted a
substantial part of Kipling's poem, that would infringe his copyright
(unless one of the Copyright Act exceptions applied). Typically in such >>> circumstances, my publisher would seek permission.
Although what publishers do, perhaps out of excessive caution, isn't
really much of an indication. An example has been given here of
Sherlock Holmes keeping bees being a no-no for publishers to repeat.
See my comment elsewhere that this seems to be based on practice in USA, rather than UK.
But I doubt very much whether just that, without quoting a considerable
amount of Conan-Doyle's actual words can amount to copyright infringement.
What is "considerable"? You may be thinking of the old "skill and
labour" test previously applied by the English courts.
However, nowadays it can be as little as 11 words, depending whether
those words amount to "the author's own intellectual creation".
This comes from the CJEU Infopaq case. That was about a newspaper
clippings service, which reproduced 11-word snippets from newspaper
articles and sent them to its subscribers.
Since Brexit, the higher UK courts have had discretion to depart from
such CJEU case law, e.g. to restore the old 'skill and labour' test.
But to the best of my knowledge, so far they have shown no signs of
doing so. The "author's own intellectual creation" test is well
established in the English courts.
As an example of that, I give you the finding that 'Del Boy is a
protected literary work', meaning the character, not a book. If that
doesn't make you wince, I don't know what will.
More generally, this was based on the Sonny Bono amendment to US
copyright law, which extended the US copyright protection of some old
works to 95 years from the date of publication. Combined with an
aggressive legal stance by the Conan Doyle estate, which went
unchallenged for a long time.
When eventually challenged in the US courts, it was held that only "increments of expression" introduced in the later stories had any protection. Note that the US courts draw a subtle distinction between
an increment of *expression* (which could be protected) and an idea
(which can't).
On Wed, 1 Jan 2025 09:27:17 +0000, Norman Wells wrote...
As an example of that, I give you the finding that 'Del Boy is a
protected literary work', meaning the character, not a book. If that
doesn't make you wince, I don't know what will.
A useful link, thank you. Whether the outcome is good or bad would be a matter of opinion.
I note that it raises similar questions to the US Sherlock case I cited previously, as to the dividing line between ideas (not protected) and
the expression of those ideas (protected). Expression is not limited
just to the words used. But "the features of Del Boy relied on by
Shazam as constituting his character were precisely and objectively discernable in the OFAH scripts."
On Tue, 31 Dec 2024 17:09:35 -0000 (UTC), Jethro_uk <jethro_uk@hotmailbin.com> wrote:
[quoted text muted]
I'd say there are quite a few humans who would have trouble
understanding it - does that mean they are not intelligent?
On Thu, 02 Jan 2025 09:06:14 +0000, Martin Harran wrote:
On Tue, 31 Dec 2024 17:09:35 -0000 (UTC), Jethro_uk
<jethro_uk@hotmailbin.com> wrote:
[quoted text muted]
I'd say there are quite a few humans who would have trouble
understanding it - does that mean they are not intelligent?
You define intelligence, and I will arbitrate :)
We are told that dogs (for example) are intelligent. Yet they invariably score zero on an IQ test. Which suggests that either (a) dogs are not intelligent; or (b) that IQ tests do not in fact measure intelligence but some other nebulous quality as a proxy for intelligence are are therefore (a:2) useless; or (b:2) a waste of money.
On Thu, 02 Jan 2025 09:06:14 +0000, Martin Harran wrote:
On Tue, 31 Dec 2024 17:09:35 -0000 (UTC), Jethro_uk
<jethro_uk@hotmailbin.com> wrote:
[quoted text muted]
I'd say there are quite a few humans who would have trouble
understanding it - does that mean they are not intelligent?
You define intelligence, and I will arbitrate :)
We are told that dogs (for example) are intelligent. Yet they invariably score zero on an IQ test. Which suggests that either (a) dogs are not intelligent; or (b) that IQ tests do not in fact measure intelligence but some other nebulous quality as a proxy for intelligence are are therefore (a:2) useless; or (b:2) a waste of money.
On Thu, 02 Jan 2025 09:06:14 +0000, Martin Harran wrote:
On Tue, 31 Dec 2024 17:09:35 -0000 (UTC), Jethro_uk
<jethro_uk@hotmailbin.com> wrote:
[quoted text muted]
I'd say there are quite a few humans who would have trouble
understanding it - does that mean they are not intelligent?
You define intelligence, and I will arbitrate :)
We are told that dogs (for example) are intelligent. Yet they invariably score zero on an IQ test. Which suggests that either (a) dogs are not intelligent; or (b) that IQ tests do not in fact measure intelligence but some other nebulous quality as a proxy for intelligence are are therefore (a:2) useless; or (b:2) a waste of money.
On Tue, 31 Dec 2024 17:09:35 -0000 (UTC), Jethro_uk <jethro_uk@hotmailbin.com> wrote:
On Tue, 31 Dec 2024 16:47:17 +0000, Tim Jackson wrote:
Earlier in the thread, someone linked a Kipling poem. Ignore for the
sake of argument the fact that Kipling's copyright has expired.
If I were to summarise that poem in my own words, it would be
derivative. But since it would be my words, not Kipling's, it wouldn't
infringe his copyright.
On the other hand, if I were to write a short story which quoted a
substantial part of Kipling's poem, that would infringe his copyright
(unless one of the Copyright Act exceptions applied). Typically in such >>> circumstances, my publisher would seek permission.
The thing is you can "understand" Kiplings poem. No "AI" engine currently
could do that. Although a few could give a damn good impersonation of
someone understanding it.
I'd say there are quite a few humans who would have trouble
understanding it - does that mean they are not intelligent?
Martin Brown <'''newspam'''@nonad.co.uk> wrote:
[trimmed]
Earlier AI symbolic algebra tools found some cute results in general
relativity too but they were credited to the author(s) of the code.
More recently Google DeepMind is now as good as a silver medalist at
the International Maths Olympiad (and that is *seriously* good). I
grant you that isn't general intelligence but it is now very close to
being able to beat all but the very best mathematicians on the
planet.
Its got a better memory and doesnt need to sleep or eat but is
not necessarily deeper thinking.
https://ukmt.org.uk/deepminds-ai-achieves-breakthrough-in-
solving-international-mathematical-olympiad-problems
and at mathematics research level AI tools are getting very good now
https://www.scientificamerican.com/article/ai-will-become-
mathematicians-co-pilot/
AI already has the edge over humans for large scale chip design.
But its all algorithmic! It merely means that AI has more memory
and faster processing than the human brain.
What has AI discovered that wasnt known about before?
On 02/01/2025 11:31, Jethro_uk wrote:
On Thu, 02 Jan 2025 09:06:14 +0000, Martin Harran wrote:
On Tue, 31 Dec 2024 17:09:35 -0000 (UTC), Jethro_uk
<jethro_uk@hotmailbin.com> wrote:
[quoted text muted]
I'd say there are quite a few humans who would have trouble
understanding it - does that mean they are not intelligent?
You define intelligence, and I will arbitrate :)
We are told that dogs (for example) are intelligent. Yet they
invariably score zero on an IQ test. Which suggests that either (a)
dogs are not intelligent; or (b) that IQ tests do not in fact measure
intelligence but some other nebulous quality as a proxy for
intelligence are are therefore (a:2) useless; or (b:2) a waste of
money.
Human IQ tests assume the ability to understand written instructions
On Thu, 2 Jan 2025 11:31:41 -0000 (UTC), Jethro_uk
<jethro_uk@hotmailbin.com> wrote:
On Thu, 02 Jan 2025 09:06:14 +0000, Martin Harran wrote:
On Tue, 31 Dec 2024 17:09:35 -0000 (UTC), Jethro_uk
<jethro_uk@hotmailbin.com> wrote:
[quoted text muted]
I'd say there are quite a few humans who would have trouble
understanding it - does that mean they are not intelligent?
You define intelligence, and I will arbitrate :)
You seemed to be assessing it on the basis of being able to understand a
poem so what definition did you have in mind for that understanding?
On 13:47 31 Dec 2024, Spike said:
Martin Brown <'''newspam'''@nonad.co.uk> wrote:
[trimmed]
Earlier AI symbolic algebra tools found some cute results in general
relativity too but they were credited to the author(s) of the code.
More recently Google DeepMind is now as good as a silver medalist at
the International Maths Olympiad (and that is *seriously* good). I
grant you that isn't general intelligence but it is now very close to
being able to beat all but the very best mathematicians on the planet.
Its got a better memory and doesnt need to sleep or eat but is not
necessarily deeper thinking.
https://ukmt.org.uk/deepminds-ai-achieves-breakthrough-in-
solving-international-mathematical-olympiad-problems
and at mathematics research level AI tools are getting very good now
https://www.scientificamerican.com/article/ai-will-become-
mathematicians-co-pilot/
AI already has the edge over humans for large scale chip design.
But its all algorithmic! It merely means that AI has more memory and
faster processing than the human brain.
What has AI discovered that wasnt known about before?
Isn't one of the features of AI that it can learn from experience,
sometimes in near real time? That's more than being simply algorithmic.
On 13:47 31 Dec 2024, Spike said:
Martin Brown <'''newspam'''@nonad.co.uk> wrote:
[trimmed]
Earlier AI symbolic algebra tools found some cute results in general
relativity too but they were credited to the author(s) of the code.
More recently Google DeepMind is now as good as a silver medalist at
the International Maths Olympiad (and that is *seriously* good). I
grant you that isn't general intelligence but it is now very close to
being able to beat all but the very best mathematicians on the
planet.
It’s got a better memory and doesn’t need to sleep or eat but is
not necessarily deeper thinking.
https://ukmt.org.uk/deepminds-ai-achieves-breakthrough-in-
solving-international-mathematical-olympiad-problems
and at mathematics research level AI tools are getting very good now
https://www.scientificamerican.com/article/ai-will-become-
mathematicians-co-pilot/
AI already has the edge over humans for large scale chip design.
But it’s all algorithmic! It merely means that AI has more memory
and faster processing than the human brain.
What has AI discovered that wasn’t known about before?
Isn't one of the features of AI that it can learn from experience,
sometimes in near real time? That's more than being simply algorithmic.
On 01/01/2025 15:44, Tim Jackson wrote:
On Wed, 1 Jan 2025 09:27:17 +0000, Norman Wells wrote...
As an example of that, I give you the finding that 'Del Boy is a
protected literary work', meaning the character, not a book. If that
doesn't make you wince, I don't know what will.
A useful link, thank you. Whether the outcome is good or bad would be a
matter of opinion.
I note that it raises similar questions to the US Sherlock case I cited
previously, as to the dividing line between ideas (not protected) and
the expression of those ideas (protected). Expression is not limited
just to the words used. But "the features of Del Boy relied on by
Shazam as constituting his character were precisely and objectively
discernable in the OFAH scripts."
As is often the case, I recommend reading the full judgment, rather than
a single page summary of it as some of the nuances may be missed in the summary. For example, the case covered claims both for copyright infringement and passing off (both of which succeeded). However, not
all claims in both headings were successful (for example "The character
'Del Boy' IS NOT considered a dramatic work for the purposes of
copyright law"), as the helpful table at para [210] of the judgment summarises everything quite nicely (with links back to the relevant
section for the rationale for arriving at said conclusions).
The full judgment can be read here:
https://www.bailii.org/ew/cases/EWHC/IPEC/2022/1379.html
| Sysop: | Keyop |
|---|---|
| Location: | Huddersfield, West Yorkshire, UK |
| Users: | 498 |
| Nodes: | 16 (2 / 14) |
| Uptime: | 27:34:49 |
| Calls: | 9,830 |
| Calls today: | 9 |
| Files: | 13,761 |
| Messages: | 6,192,325 |
| Posted today: | 1 |